managed flood storage option for selected levees along the lower … · 2006. 8. 24. · managed...
TRANSCRIPT
Contract Report 590
Managed Flood Storage Option for Selected Levees along the Lower Illinois River
for Enhancing Flood Protection, Agriculture, Wetlands, and Recreation
First Report: Stage and Flood Frequencies and the Mississippi Backwater Effects
by Krishan P. Singh
Office of Surface Water Resources: Systems, Information, and GIS
Prepared for the Office of Water Resources Management
February 1996
Illinois State Water Survey Hydrology Division Champaign, Illinois
A Division of the Illinois Department of Natural Resources
Managed Flood Storage Option for Selected Levees along the Lower Illinois River
for Enhancing Flood Protection, Agriculture, Wetlands, and Recreation
First Report: Stage and Flood Frequencies and the Mississippi Backwater Effects
by Krishan P. Singh, Principal Scientist Office of Surface Water Resources: Systems, Information, and GIS
Illinois State Water Survey 2204 Griffith Drive
Champaign, Illinois 61820
February 1996
ISSN 0733-3927
This report was printed on recycled and recyclable papers.
CONTENTS Page
Introduction 1 Managed Flood Storage Option 5 Acknowledgments 7
A General Frequency Program 10 Rationale for Mixed Distributions 10 Mixed Distribution 11 Outlier/Inlier Detection and Modification 12 Kurtosis Correction 16 Distributions Used 16
Normal Distribution 16 Log-Pearson Type HI Distribution or LP3 16 Mixed Distribution 19
New Flood Frequency Methodology 19 The Flow Chart 19 Examples 22 General Frequency Method 30
Water Stages and Elevations in Alton Pool, Illinois River 33 Annual Maximum Water Elevations 33
WE Rankings for Stations in Alton Pool 41 Stage Frequency Analysis 45 T-Year River Water Elevations 56 Grafton Stages and Meredosia Flood Peaks 56 Highest Observed Water Surface Profiles, Alton Pool, Illinois River 60 Observed Stage Hydrographs for 1943 and 1993 Events 62
Flood Frequency Analyses for Major Tributaries 67 Design T-Year Floods 67
Mackinaw River near Green Valley 79 Spoon River at Seville 82 Sangamon River at Oakford 84 La Moine River at Ripley , 84 Macoupin Creek near Kane 86
Flood Peak versus T-curves 88
Floods and Stages: Illinois River below Marseilles 91 Flood Frequency and Design Floods 91
Flood Frequency Analyses 93 Design Floods 93
Stage Frequency and Design Stages 105 Concurrency of Tributary and Illinois River Flood Peaks 113
iii
Flood of May 1943 120 Flood of March 1985 122 Flood of December 1982 122 Flood of April 1979 125 Flood of June 1974 127
Comparisons and Conclusions 129 T-Year Water Elevations in Alton Pool, Illinois River 129 T-Year Flood Peaks in Major Tributaries 131 T-Year Flood Peaks at Illinois River Gaging Stations 131 T-Year Water Elevations at Marseilles, Kingston Mines, and Meredosia 133 Conclusions 135
References 137
iv
List of Figures
Figure 1. The Illinois River and navigation locks and dams 2
Figure 2. Schematics of flood storage option 6
Figure 3. 100-year flood and stage hydrographs under different conditions 8
Figure 4. Distribution of departures for outliers and inliers 14
Figure 5. Successive testing for outliers/inliers and their modification 14
Figure 6. Flow chart for computer program 20
Figure 7. Sangamon River near Oakford, fitted flood probability curves 26
Figure 8. Mississippi River at Grafton, fitted stage probability curves 31
Figure 9. Annual maximum water elevations, Illinois River at Meredosia 34
Figure 10. Annual maximum water elevations, Mississippi River at Grafton 35
Figure 11. Annual maximum water elevations at Meredosia and Grafton, 1941 -1993 42
Figure 12. Fitted mixed and log-Pearson HI distributions to observed annual 48 annual maximum stages, Illinois River at Meredosia, 1941-1993
Figure 13. Fitted mixed and log-Pearson III distributions to observed annual 49 maximum stages, Illinois River at Valley City, 1941-1993
Figure 14. Fitted mixed and log-Pearson III distributions to observed annual 50 maximum stages, Illinois River at Florence, 1942-1993
Figure 15. Fitted mixed and log-Pearson III distributions to observed annual 51 maximum stages, Illinois River at Pearl, 1942-1993
Figure 16. Fitted mixed and log-Pearson HI distributions to observed annual 52 maximum stages, Illinois River at Hardin, 1941-1993
Figure 17. Fitted mixed and log-Pearson IE distributions to observed annual 53 maximum stages, Mississippi River at Grafton, 1941-1993
Figure 18. T-year maximum water elevation profiles for Illinois River from Meredosia 57 to Grafton (T is average recurrence interval in years)
v
Figure 19. Meredosia flood peak ranks versus Grafton maximum water elevation ranks, 58 using 1941-1993 data
Figure 20. Highest observed water surface profiles, Meredosia to Grafton, 1941 -1993 61 (WE3 is WE with rank 3, Q3 is flood peak with rank 3, etc.)
Figure 21. Water elevation versus time curves at Meredosia, Valley City, Florence, 63 Pearl, Hardin, and Grafton during May 17-31, 1943
Figure 22. Water elevation versus time curves at Meredosia, Valley City, Florence, 65 Pearl, Hardin, and Grafton during July 25 to August 8, 1993
Figure 23. Fitted MD and LP3s distributions to observed annual maximum floods: 74 Mackinaw River near Green Valley, 1941-1993
Figure 24. Fitted MD and LP3s distributions to observed annual maximum floods: 75 Spoon River at Seville, 1941-1993
Figure 25. Fitted MD and LP3s distributions to observed annual maximum floods: 76 Sangamon River near Oakford, 1941 -1993
Figure 26. Fitted MD and LP3s distributions to observed annual maximum floods: 77 La Moine River at Ripley, 1941-1993
Figure 27. Fitted MD and LP3s distributions to observed annual maximum floods: 78 Macoupin Creek near Kane, 1941-1993
Figure 28. Fitted MD and LP3s distributions to observed annual maximum floods: 80 Mackinaw River near Green Valley, 1922-1951
Figure 29. 2- to 1000-year flood peaks for various periods: Mackinaw River 81
Figure 30. 2- to 1000-year flood peaks for various periods: Spoon River at Seville 83
Figure 31. 2- to 1000-year flood peaks for various periods: Sangamon River 85 near Oakford
Figure 32. 2- to 1000-year flood peaks for various periods: La Moine River at Ripley 87
Figure 33. 2- to 1000-year flood peaks for various periods: Macoupin Creek near Kane 89
Figure 34. Flood peak versus T curves for Mackinaw, Spoon, Sangamon, and 90 La Moine Rivers, and Macoupin Creek
vi
Figure 35. Fitted MD and LP3s distribution to annual maximum floods, Illinois River 99 at Marseilles, 1969-1993
Figure 36. Fitted MD and LP3s distribution to annual maximum floods, Illinois River 100 at Kingston Mines, 1969-1993
Figure 37. Fitted MD and LP3s distribution to annual maximum floods, Illinois River 101 at Meredosia, 1969-1993
Figure 38. Fitted MD and LP3s distribution to annual maximum floods, Illinois River 102 at Marseilles, 1941-1993
Figure 39 . Fitted MD and LP3s distribution to annual maximum floods, Illinois River 103 at Kingston Mines, 1941-1993
Figure 40. Fitted MD and LP3s distribution to annual maximum floods, Illinois River 104 at Meredosia, 1941-1993
Figure 41. T-year flood peaks for Illinois River at Marseilles, Kingston Mines, 106 and Meredosia
Figure 42. T-year stages for Illinois River at Marseilles, Kingston Mines, 111 and Meredosia
Figure 43. T-year water elevations for Illinois River at Marseilles, Kingston Mines, 112 and Meredosia
Figure 44. Flood hydrographs at Illinois River and tributary gaging stations, 121 May 1943 flood
Figure 45. Flood hydrographs at Illinois River and tributary gaging stations, 123 March 1985 flood
Figure 46. Flood hydrographs at Illinois River and tributary gaging stations, 124 Dec. 1982 flood
Figure 47. Flood hydrographs at Illinois River and tributary gaging stations, 126 April 1979 flood
Figure 48. Rood hydrographs at Illinois River and tributary gaging stations, 128 June 1974 flood
vii
List of Tables
Table 1. Levee and Drainage Districts along the Illinois River below Peoria Lake 4 to Grafton, Illinois
Table 2. Test Values of Outlier and Inlier Departures 15
Table 3. Values of ZT for Various Values of β and T 17
Table 4. Flood Frequency Analysis for Sangamon River near Oakford, 23 Station Number 05583000, 1941-1993
Table 5. Flood Frequency Analysis for Sangamon River near Oakford, 24 Station Number 05583000, 1941-1993
Table 6. Stage Frequency Analysis for Mississippi River at Grafton, 27 Station Number 05585500, 1941-1993
Table 7. Stage Frequency Analysis for Mississippi River at Grafton, 28 Station Number 05585500, 1941-1993
Table 8. Annual Maximum Daily Water Elevations and Dates at Stage Stations 36 in the Alton Pool, Illinois River (1941-1993)
Table 9. Ranked Annual Maximum Water Elevations and Floods at Meredosia and 43 Related Water Elevation Ranks at Other Stations in Alton Pool, Illinois River
Table 10. Parameters for Fitted Mixed and Log-Pearson HI Distributions 46
Table 11. Various Recurrence-Interval Stages in Feet at Stage Recording Stations 54 in Alton Pool, Illinois River
Table 12. Observed Maximum Stages, Dates, and Fitted Recurrence Intervals 55 for Stations in Alton Pool, Illinois River
Table 13. Various Recurrence-Interval River Water Elevations with MD at 55 Stage Stations in Alton Pool, Illinois River
Table 14. Relevant Data for Major Tributaries to Illinois River below Peoria, Illinois 68
Table 15. Fitted Mixed Distribution (MD) and Log-Pearson III (LP3s) 69 Distribution Parameters
Table 16. Flood Frequency Analysis for Major Tributaries to Illinois River 70 below Peoria, Illinois
viii
Table 17. Record Periods, Outlier/Inlier Modifications, and Maximum Flood Peaks 73
Table 18. Ranked Peak Floods at Marseilles, Kingston Mines, and 92 Meredosia (1941-1993)
Table 19. Fitted Mixed Distribution (MD) and Log-Pearson III (LP3s) Distribution 94 Parameters for Peak Floods
Table 20. Flood Frequency Analysis for Illinois River at Marseilles, Kingston Mines, 95 and Meredosia
Table 21. Record Periods, Outlier/Inlier Modifications, and Maximum Flood Peaks 98
Table 22. Fitted Mixed Distribution (MD) and Log-Pearson III Distribution (LP3s) 107 Parameters for Peak Stages
Table 23. Stage Frequency Analysis for Illinois River at Marseilles, Kingston Mines, 108 and Meredosia
Table 24. Relevant Data for the Illinois River and Major Tributaries below Marseilles 114
Table 25. Peak Floods in the Illinois River and Major Tributaries below Marseilles 115
Table 26. Relevant Information for Top Five Floods: Marseilles to Meredosia 118
Table 27. Comparison of T-year Water Elevations in this Study and from USCOE 130 Report for Alton Pool
Table 28. Comparison of T-year Floods in Major Tributaries to the Illinois River 132
Table 29. Comparison of T-year Hoods at Marseilles, Kingston Mines, and Meredosia 134
Table 30. Comparison of T-year Water Elevations at Marseilles, Kingston Mines, 134 and Meredosia
ix
Managed Flood Storage Option for Selected Levees along the Lower Illinois River for Enhancing Flood Protection, Agriculture, Wetlands, and Recreation
First Report: Stage and Flood Frequencies and the Mississippi Backwater Effects
INTRODUCTION
This report is the first of three reports that are planned for a three-year study being
supported by the Office of Water Resources Management, Illinois Department of Natural
Resources (previously Division of Water Resources, Illinois Department of Transportation). The
study is entitled Managed Flood Storage Option for Selected Levees along the Lower Illinois
River for Enhancing Flood Protection, Agriculture, Wetlands, and Recreation. The main goal is
to examine the hydrologic and economic benefits of modifying some at-risk levees and drainage
districts along the lower Illinois River to provide managed flood storage (as well as wetland and
recreation functions), and to increase flood protection for other levees and districts. The study
involves validation of UNET (one-dimensional unsteady flow through a full network of open
channels) model (Barkau, 1993) application to the Illinois River, simulating flood profiles under
various scenarios of tributary and main river flows, identifying levees at risk as possible
candidates for modification to the flood storage option, determining a minimum set of such
levees to obtain desirable protection against failure by high flood for other levees, and some
investigation of the land use behind modified levees to maximize a mix of such uses as
agriculture, recreation, wetlands, etc. The first step was to study the flood and stage frequencies
applied to stations along the Illinois River and its major tributaries, as well as to understand the
development of high floods in the Illinois River vis-a-vis the probability of high floods in the
tributaries. This will be followed by model verification and validation, simulations, and other
analyses as envisaged under the general objectives.
The Illinois River has been the focus of intensive study for more than 60 years, largely
because of alterations caused by Lake Michigan inflows, wastewater effluents from the Chicago
metropolitan area, and silting of backwater lakes and Peoria Lake, which significantly affected
ecosystems, levee and drainage districts mostly below Peoria, and conversion of fast-flowing
river to a navigable river with relatively flat pools (Figure 1). The levees and drainage districts
removed about 200,000 acres of floodplain area. Flood levels kept on steadily rising with
1
Figure 1. The Illinois River and navigation locks and dams
2
construction of new levees mostly completed in the 1920s. Table 1 lists the levee and drainage
districts between Peoria and Grafton. Seven major locks and dams built in the 1930s as part of
the Illinois Waterway provided enough surcharge storage in the pools to moderate the high flows
and levels.
Flood magnitudes have considerably increased at Marseilles, Kingston Mines, and
Meredosia since 1970. The top five floods (ranks 1 - 5) at these stations during the 1941-1993
period (after the completion of locks and dams on the Illinois River) occurred in Water Years
1983, 1957, 1970, 1981, and 1991 at Marseilles; 1983, 1943, 1985, 1982, and 1970 at Kingston
Mines; and 1943, 1985, 1983, 1979, and 1974 at Meredosia. Thus, four out of the top five floods
in the 1941-1993 period occurred during 1970-1993. Though one cannot surmise that this
upward trend will continue, some notice of this trend needs to be incorporated in flood and stage
frequency analyses. Some of the increase in flood peaks is attributed to a trend of increasing
precipitation in the upper half of the Illinois River basin (Singh and Ramamurthy, 1990).
The Illinois River is a relatively large, managed river, and the applicability of particular
probability distributions needs to be checked and validated. In the river reach below Peoria to
Grafton, the LaGrange lock and dam creates LaGrange Pool, and the Alton lock and dam in the
Mississippi River controls the Alton Pool in the Elinois River. The latter involves interaction
With floods and stages in the Mississippi River. These pools do attenuate to some extent the
flood flow in these reaches, but at high flow this effect decreases because of the limited extra
storage provided by these pools. Under such conditions the annual floods and stages may be
analyzed using usual frequency distributions.
The log-Pearson HI (LP3) distribution is being widely used, with some correction for
skew value when the sample size is not very large. Use of regional skew to develop a weighted
skew does cause problems when sample skew is a significantly positive value. Singh (1968,
1980b, 1983) and Singh and Nakashima (1981) have developed a versatile frequency analysis
method that detects and modifies any outliers and inliers at significance levels of 0.01, 0.05, 0.10,
0.20, 0.30, and 0.40 in various size samples and develops estimates of various recurrence interval
floods using power-transformed normal distribution (PT), LP3, and mixed distribution (MD). An
analysis of these flood estimates in terms of increase in peak values with increase in recurrence
interval, consistency of results, and relative fitting of the distribution curves to the observed flood
3
Table 1. Levee and Drainage Districts along the Illinois River below Peoria Lake to Grafton, Illinois
Levee and Drainage District Upstream river mile Downstream river mile
Pekin & Lamarsh 155.0 149.7 Spring Lake 147.2 134.2 Benner Special 146.0 138.0 East Liverpool 132.0 129.0 Liverpool 129.0 127.0 Thompson Lake 127.0 119.4 Lacey, Langellier, W. Matanzas & Kerton 119.4 111.8 Seahorn 111.8 107.0 Big Lake 107.0 102.0 Kelly Lake 102.0 97.0 Coal Creek 91.2 85.0 S. Beardstown & Valley 87.7 79.0 Crane Creek 85.1 83.8 Meredosia Lake & Willow Creek 79.0 72.7 Little Creek 78.3 75.1 McGee Creek 75.0 67.1 Valley City 66.2 62.5 Mauvaise Terre 65.8 63.4 Scott County 63.1 56.7 Big Swan 56.5 50.1 Hillview 50.0 43.2 Hartwell 43.1 38.2 Keach 38.0 32.8 Eldred & Spanky 32.4 23.8 Nutwood 23.6 15.1
4
data, can be carried out. Any trend of increasing flood peaks and stages can be examined by
doing frequency analysis on split samples.
The Illinois River flood stage elevations in the Alton Pool (from Meredosia to Grafton)
are governed by a mix of varying circumstances: 1) the backwater effects caused by the
Mississippi River levels raise water stages in the Illinois River — the higher the Mississippi River
level, the greater the distance upstream the backwater effect raises the water elevation in the
Illinois River, 2) the greater the flood peak at Meredosia, the greater the distance downstream not
experiencing Mississippi River backwater effects, 3) a combination of rather low flood at
Meredosia and very high water stage at Grafton can push the backwater effects all the way up to
Meredosia lock and dam, and 4) a combination of a very high flood at Meredosia and very low
Mississippi River water elevation will restrict backwater effects in the Illinois to a small distance
upstream of Grafton.
Managed Flood Storage Option
A preliminary study (Singh, 1991) indicates that an overall economic benefit may be
achieved by conversion of some at-risk, selected levee districts to a managed flood storage
function. The effects of a simple floodplain and managed storage on the peak stage and
discharge of a 100-year flood at the Kingston Mines gage were estimated in a preliminary
hydrologic analysis. The levee was considered to have crown elevation of 456.0 feet above mean
sea level (feet-msl) or about the l00-year flood stage level of 455.8 feet-msl used in the
preliminary study. This leaves a freeboard of only 0.2 feet instead of the usual 2 to 3 feet. A cut
500 feet long and 7 feet in depth was assumed to act as a broad-crested weir over which
floodwaters would flow into the levee interior of 10,000 acres when Illinois River water
elevations exceeded 449.0 feet-msl. Flow over this cut or fixed overflow section in the levee was
simulated using a broad-crested weir flow equation and the volume of water that enters the area
behind the levee (or levee interior) was subtracted from the flow in the river. Results suggest
that a large amount of flood storage, such as that provided by inundating an entire levee district,
can (reduce the peak stage at the levee by 2.3 feet (Figure 2) only if the flow into the levee interior
is controlled or "managed" so that it occurs near the peak of the flood hydrograph. When the
Illinois River is at peak stage of 453.5 feet-msl, water in the managed storage is at 447.2 feet-msl
5
100-Year Flood Stages
Figure 2. Schematics of flood storage option
6
and it rises to 451.4 feet-msl as the river level falls to that level. If an entire section of a levee is
removed, then the inflow into the area, which had been protected by the levee, begins early
during the flood, similar to the condition if there was no levee at all. Under this "no-levee"
condition, much of the floodplain storage is filled early in the flood so it is not available to
reduce the peak stage. Consequently, the reduction in the 100-year flood elevation is only 0.3
feet. Under the flood-storage option, the stored floodwater will begin flowing back into the river
through bottom outlets after the river stage starts to recede. Unsteady modeling of the river flow
is necessary to address flow dynamics and better estimation of the flood storage effect on river
flow. The effect of managed flood storage would be to reduce the stage hydrograph only at high
discharges, greatly lowering the peak stage (Figure 3) as well as the flood peak. The optimal
flood stage at which the planned overflow occurs would have to be determined from various
scenarios of at-risk levee conversions. The proposed research will analyze the effects of simple
floodplain and managed flood storage by UNET model simulations of the flood stages resulting
from varying flow conditions.
Additional benefits may be obtained by using the land behind storage-option levees for
other functions (agriculture, wetlands, recreation, etc.). Although the primary function proposed
for analysis is that of flood storage, it is likely that the levee overflow structures may be designed
at a level where the flood overflow into the levee interior may occur with an average frequency
of once every 10 to 20 years. Under these conditions, it will be feasible to use the lands for other
functions for most years.
Acknowledgments
The study is jointly supported by the Office of Water Resource Management, Illinois
Department of Natural Resources (previously Division of Water Resources, Illinois Department
of Transportation) and the Illinois State Water Survey, Illinois Department of Natural Resources
(previously Illinois State Water Survey, Illinois Department of Energy and Natural Resources).
Gary Clark, Office of Water Resource Management, is serving in a liaison capacity during the
entire course of this study. The U.S. Army Corps of Engineers District offices in St. Louis,
Missouri, and Rock Island, Dlinois, provided some of the basic data used in this study. Computer
runs for flood and stage frequency analyses were carried out by graduate research assistant
7
Figure 3. 100-year flood and stage hydrographs under different conditions
8
Zhaoyun Xing, who is studying for his Ph.D. in computer sciences at the University of Illinois.
Kathleen Brown typed the manuscript, Linda Hascall assisted with the graphics, and Eva
Kingston edited the report.
9
A GENERAL FREQUENCY PROGRAM
Many observed annual flood series exhibit reverse curvatures when plotted on lognormal
probability paper. The most commonly used distributions in flood frequency analyses are the
Pearson and log-Pearson, normal and lognormal, and Gumbel (extreme value) and log-Gumbel.
None of these distributions, however, fits an observed flood series with reverse curvature. The
occurrence of these curvatures may be attributed to seasonal variation in flood-producing storms,
dominance of flow within the channel or floodplain, and variability in antecedent basin soil
moisture and cover conditions. A mixed distribution model is required to analyze such flood
series because of the mixed population of floods.
Existence of any significant outliers, inliers, or both in the flood series can lead to
substantial bias in computed distribution parameters and high flood estimates. Such outliers and
inliers need to be detected and modified. Only one detection and modification methodology
currently exists (Singh, 1987).
This versatile flood frequency methodology uses a mixed distribution (MD) model to
simulate the observed flood series, with objective detection and modification of any outliers and
inliers at various significance levels. This methodology has been computerized and tested on
hundreds of flood series from various parts of the world.
Rationale for Mixed Distributions
Moran (1959) calls the expression a1 p1(x) + . . . . + ak pk(x) a mixture of probability
distributions p1(x), , pk(x), if the a1 . . . . , ak are non-negative and a1 + . . . . + ak = 1.
Hald (1952) defines a population formed by two populations in a given proportion as a
heterogeneous population. From careful analyses of the lognormal probability plots of observed
annual flood series at many streamgaging stations in various parts of the world, Singh (1968)
concludes that reverse curvatures in these plots are caused by the heterogeneity of the population
of floods.
The magnitude of a flood peak depends largely on storm and basin characteristics. Storm
characteristics of interest are the type of storm and intensity and duration of the storm. Rainfall
may be caused by hurricanes, thunderstorms, and frontal or air-mass storms. High intensity and
10
short duration storms usually cause much higher peak flow than low intensity and long duration
storms with the same total precipitation. Basin characteristics mainly include relative dominance
of flow within the channel or floodplain, the antecedent soil moisture condition, and the vegetal
cover.
The effect of channel versus floodplain flow is explained as follows. Bankfull discharge
in a river corresponds to about a 2-year or median flood. With an increase in flow, the water
spreads over the floodplain. For low depths of inundation, the mean velocity of the composite
flow section is much lower than at the bankfull discharge. The mean velocity slowly increases
with increase in depth of flooding and may surpass the bankfull flow velocity. This addition of a
new storage element can lead to flattening and subsequent rise of the flood probability curve at a
different slope than for the floods within the channel. Conversion of storm rainfall or runoff is
largely affected by the antecedent soil moisture condition and vegetal cover.
The magnitude of annual flood peaks depends on a number of factors that vary within a
season and from season to season. The interaction between the distributions of these pertinent
factors may produce a flood series resembling a conventional distribution shape or one exhibiting
marked reverse curvature to be dealt with by the mixed-distribution concept. Mixed distributions
have been considered in terms of rainfall and snowmelt floods (Waylen and Woo, 1982) and
floods caused by hurricanes and other storms (Canterford and Pierrehumbert, 1977).
Mixed Distribution
The mixed distribution (MD) model (Singh, 1968; Singh and Sinclair, 1972; Singh, 1974)
considers the observed annual maximum floods (or their logarithms) to belong to two
populations with means µ1, and µ2, variances σ12 and σ2
2, and relative weights a and \-a:
in which p is the probability of being equal to or less than x, and x = log Q where Q is the annual
flood. The five parameters (µ1, µ2, σ1, σ2, and a) are linked to statistics: mean, standard
11
deviation, and skew of observed annual flood series (Cohen, 1967). The MD model considers an
observed flood series as essentially composed of two component distributions. Use of three or
more component distributions increases the complexity of the problem and makes it intractable.
The values of the parameters can be obtained with a nonlinear programming algorithm
(Singh and Nakashima, 1981). The nonlinear objective function is minimization of ∑ | ∆Z |
where AZ equals the difference between standard deviate corresponding to the observed
probability equal to (m - 0.38)/(n + 0.24) and that fitted corresponding to p from the mixed-
distribution equation; m is the ranked order for the flood series and n is the sample size. The
following constraints apply.
Outlier/Inlier Detection and Modification
Barnett and Lewis (1978) define an outlier in a set of data as an observation or a subset of
observations that appears to be inconsistent with the remainder of that set of data. When the
values of the highest observed floods of an annual flood series are much higher or lower than
expected, these values are designated as outliers and inliers, respectively. When the values of the
lowest observed floods are much higher or lower than expected, these are designated as inliers
and outliers, respectively. Analyses of storms causing outliers at the high end and of droughts
causing outliers at the low end can provide a physical rationale for the presence of outliers and
inliers.
A literature search showed availability of statistical tests for checking outliers at 0.01 and
0.05 levels, but no tests for inliers. Singh and Nakashima (1981) developed an objective
methodology for successive detection and modification of any outliers and inliers at both ends of
the flood spectrum at 0.01, 0.05, 0.10, 0.20, 0.30, and 0.40 levels of significance, from
12
experiments on millions of normally distributed numbers. The developed test statistic is termed
a departure, which is given by
where Z is the theoretical standard normal deviate and Zs is the sample standardized deviate
corresponding to the plotting position, p
where p is the probability of nonexceedance, m is the rank order for the flood series ranked from
low to high, and a = 0.38 (Blom, 1958). Figure 4 shows the values of the departures for the five
highest and five lowest floods in an annual flood series at various probability levels. In this
figure the five highest floods are ranked from high to low, and the five lowest floods are ranked
from low to high (designated by 1,2,3,4, and 5, respectively). Table 2 provides test values of
outlier and inlier departures determined at 23 probabilities for sample sizes 10, 15, 20, 25, 30, 40,
50, 60, 75 and 100. Usually the number of outliers and inliers at the high and low end of the
flood spectrum increases with the sample size.
The observed annual flood series needs to be transformed to resemble a series distributed
asN(µ, σ2). This is achieved with the power transformation (Box and Cox, 1964):
and
where Q is the annual flood, and λ is the transformation parameter. The λ, can be obtained with
the maximum log-likelihood method (Singh, 1980a):
and
The skew of the y series is very close to zero but the kurtosis may be different than for a normal
distribution. Singh and Nakashima (1981) provide adjustment values for various values of
13
Figure 4. Distribution of departures for outliers and inliers
Figure 5. Successive testing for outliers/inliers and their modification
14
Table 2. Test Values of Outlier and Inlier Departures
Test values of departures Outlier/ Low 1 Low 2 Low 3 Low 4 Low 5
Window p inlier 15-100 20-100 25-100 30-100 40-100
1 <0.01 Inlier <-0.689 <-0.495 <-0.412 <-0.363 <-0.327 >0.99 Outlier >1.029 >0.643 >0.498 >0.418 >0.368
2 <0.05 Inlier <-0.532 <-0.369 <-0.303 <-0.264 <-0.237 >0.95 Outlier >0.681 >0.421 >0.337 >0.285 >0.253
3 <0.10 Inlier <-0.441 <-0.299 <-0.243 <-0.211 <-0.188 >0.90 Outlier >0.503 >0.321 >0.254 >0.217 >0.193
4 <0.20 Inlier <-0.318 <-0.209 <-0.167 <-0.143 <-0.127 >0.80 Outlier >0.297 >0.197 >0.159 >0.137 >0.123
5 <0.30 Inlier <-0.221 <-0.141 <-0.110 <-0.093 <-0.082 >0.70 Outlier >0.161 >0.112 >0.092 >0.081 >0.073
6 <0.40 Inlier <-0.132 <-0.080 <-0.060 <-0.050 <-0.043 >0.60 Outlier >0.052 >0.043 >0.037 >0.034 >0.032
Outlier/ High 1 High 2 High 3 High 4 High 5 p inlier 15-100 20-100 25-100 30-100 40-100
1 <0.01 Inlier <-1.054 <-0.654 <-0.511 <-0.429 <-0.377 >0.99 Outlier >0.679 >0.488 >0.407 >0.358 >0.323
2 <0.05 Inlier <-0.683 <-0.433 <-0.341 <-0.290 <-0.256 >0.95 Outlier >0.529 >0.369 >0.300 >0.263 >0.235
3 <0.10 Inlier <-0.500 <-0.322 <-0.258 <-0.220 <-0.195 >0.90 Outlier >0.438 >0.299 >0.241 >0.209 >0.186
4 <0.20 Inlier <-0.295 <-0.197 <-0.161 <-0.139 <-0.124 >0.80 Outlier >0.317 >0.209 >0.166 >0.143 >0.126
5 <0.30 Inlier <-0.159 <-0.112 <-0.094 <-0.082 <-0.074 >0.70 Outlier >0.221 >0.140 >0.110 >0.093 >0.082
6 <0.40 Inlier <-0.051 <-0.043 <-0.039 <-0.035 <-0.032 >0.60 Outlier >0.132 >0.079 >0.060 >0.050 >0.043
Notes: 15-100,..., and 40-100 denote the range of sample size n in years, and p is the probability or significance level; windows 1,2,..., 6 refer to significance levels of 0.01, 0.05,. . . , 0.40 used for detection of outliers and inliers.
15
kurtosis for symmetric distributions. In the case of the asymmetrical y series, these adjustment
factors do not yield the best results.
The detection and modification procedure begins from window 1 or significance or
probability level 0.01, and any outliers/inliers detected are modified at that level (Figure 5). The
resulting detransformed series is analyzed to test the departures, and the procedure is repeated, if
necessary, to ensure that there are no outliers/inliers at the 0.01 significance level. By following
the procedure sequentially from one level to the next, desired distribution statistics and floods are
computed before moving to the next level. If no outliers/inliers are detected at a level, no
modifications are done for that level. Windows 1, 2, 3, 4, 5, and 6 correspond to significance or
probability levels (SL) of 0.01, 0.05, 0.10, 0.20, 0.30, and 0.40, or their complements,
respectively.
Kurtosis Correction
The power-transformed series, y, has a skew very close to zero but the kurtosis, kt, may
not equal 3 as for a normal distribution. The kurtosis correction factors were developed
following the procedure outlined by Box and Tiao (1973). Table 3 provides values of standard
deviates with kurtosis correction. Parameter (3 is related to kurtosis, kt, by the expression:
Distributions Used
Normal Distribution. The power-transformed series is considered a normal distribution,
N(x,s2), in which x is the mean and s is the standard deviation of y series. The estimate for a T-
year flood is obtained from:
where zT is with β = 0 without kurtosis correction or with p corresponding to kt for the y series.
The yT is then transformed to QT with inverse transformation:
Log-Pearson Type III Distribution or LP3. The power-transformed series is
retransformed to the Q series after any detection and modification of outliers and inliers. The Q
16
Table 3. Values of ZT for Various Values of β and T
Values of ZTfor recurrence interval, T β 10 25 50 100 500 1000
-1.00 1.386 1.593 1.663 1.697 1.725 1.729 -0.95 1.384 1.592 1.665 1.708 1.762 1.777 -0.90 1.378 1.594 1.679 1.736 1.817 1.841 0.85 1.372 1.600 1.699 1.769 1.875 1.908
-0.80 1.366 1.608 1.721 1.803 1.935 1.977
-0.75 1.360 1.618 1.744 1.839 1.996 2.047 -0.70 1.355 1.629 1.768 1.875 2.056 2.117 -0.65 0.350 1.640 1.792 1.911 2.117 2.187 -0.60 1.345 1.651 1.815 1.946 2.178 2.257 -0.55 1.340 1.661 1.838 1.982 2.238 2.328
-0.50 1.335 1.672 1.861 2.016 2.298 2.398 -0.45 1.330 1.682 1.883 2.050 2.358 2.468 -0.40 1.326 1.691 1.904 2.083 2.418 2.538 -0.35 1.321 1.700 1.925 2.116 2.477 2.608 -0.30 1.315 1.709 1.945 2.148 2.535 2.677
-0.25 1.310 1.717 1.965 2.179 2.594 2.747 -0.20 1.305 1.725 1.984 2.210 2.651 2.816 -0.15 1.299 1.732 2.002 2.240 2.709 2.885 -0.10 1.293 1.739 2.020 2.269 2.766 2.954 -0.05 1.288 1.745 2.037 2.298 2.822 3.022
0.00 1.282 1.751 2.054 2.326 2.878 3.090
0.05 1.275 1.756 2.070 2.354 2.934 3.158 0.10 1.269 1.761 2.085 2.381 2.989 3.226 0.15 1.263 1.765 2.100 2.407 3.044 3.293 0.20 1.256 1.770 2.114 2.433 3.098 3.361 0.25 1.249 1.773 2.128 2.458 3.152 3.428
0.30 1.243 1.776 2.141 2.482 3.205 3.494 0.35 1.236 1.779 2.554 2.506 3.258 3.561 0.40 1.229 1.782 2.166 2.529 3.311 3.627 0.45 1.222 1.784 2.178 2.552 3.363 3.692 0.50 1.214 1.786 2.189 2.574 3.414 3.758
0.55 1.207 1.787 2.200 2.596 3.465 3.823 0.60 1.200 1.788 2.210 2.617 3.516 3.888 0.65 1.192 1.789 2.220 2.637 3.566 3.952
17
Table 3. Concluded
Values of ZT for recurrence interval, T β 10 25 50 100 500 1000
0.70 1.185 1.789 2.229 2.657 3.616 4.016 0.75 1.177 1.790 2.238 2.677 3.665 4.080
0.80 1.169 1.789 2.247 2.695 3.714 4.143 0.85 1.162 1.789 2.255 2.714 3.762 4.206 0.90 1.154 1.788 2.262 2.732 3.810 4.269 0.95 1.146 1.787 2.269 2.749 3.857 4.331 1.00 1.138 1.786 2.276 2.766 3.904 4.393
18
series is analyzed as an LP3 distribution (U.S. Water Resources Council, 1973) and the T-year
flood estimate is obtained with the sample skew gs as well as the weighted skew gw. The
weighted skew gw is obtained from
gW = g s W+(1 -W)gr
where w equals (n - 25)/75 and lies between 0 and 1, and gr in the regional skew.
Mixed Distribution. The mixed distribution concept considers logarithms of annual
floods to belong to two populations with means µ1 and µ2, variances σ12 and σ2
2, and relative
weights a and 1 - a. Mixed distribution is a versatile distribution and can match most of the
observed flood distribution shapes with proper values of a, µ1, µ2, σ1, and σ2. Kurtosis correction
is valid only if the observed or power-transformed distribution is symmetrical. However, the
mixed distribution allows for various combinations of skew, kurtosis, and asymmetries observed
even after power transformation.
New Flood Frequency Methodology
Table 2 and Figures 4 and 5 clarify the concept of levels and windows. For the highest
flood, the outlier H1 lies in window 1 if departure A < -1.054, in window 2 if -1.054 < A <
-0.683, and so on for windows 3-6. If some outliers and/or inliers are detected in window 1, their
departures are modified to respective values at level 1, and the procedure is followed sequentially
from one window to the next. If no outliers and/or inliers are detected in a particular window, no
modification is needed, and the program moves to the next window after developing and printing
distribution statistics and flood estimates.
The Flow Chart
The detection and modification of outliers and inliers, as well as flood frequency analysis,
follows the flow chart given in Figure 6. Some relevant explanations to clarify the methodology
and the computer program are given below. The sequence numbers correspond to the numbers
attached to various boxes in the flow chart.
19
Figure 6. Flow chart for computer program
(1) Number of low as well as high floods, NO, that need to be checked to determine if
they are outliers/inliers, can be provided as input information or computed from NO = [n/10]
where NO = 5 for n > 50; n is the sample size of floods.
(2) Standard normal deviates from NO floods at both high and low end of the ranked
flood series are obtained by converting p to z with a p-z subroutine, assuming a normal
distribution:
The value of a is obtained from the previous information generated during the development of
departure test statistics.
a values for the 5 highest and lowest ranks
n 1 2 3 4 5
10 0.425 0.474 0.492 0.506 0.511 15 0.414 0.464 0.485 0.498 0.506 20 0.408 0.455 0.478 0.491 0.501 25 0.406 0.448 0.472 0.486 0.496 30 0.404 0.443 0.467 0.481 0.491 40 0.403 0.440 0.459 0.473 0.482 50 0.403 0.440 0.454 0.467 0.475 60 0.403 0.440 0.451 0.462 0.469 75 0.403 0.440 0.450 0.458 0.463
100 0.403 0.440 0.450 0.456 0.460
(3) The parameter A, is computed.
(4) The given Q series is transformed to a y series.
(5) The y series is standardized to a Y series with:
where y and ys are the mean and standard deviation of the y series.
(6) The departures, ∆ m , for the NO values at the low end as well as at the high end are
obtained from:
(7) Outliers and inliers, if any, are detected in each of the six windows according to the
six levels, with departure values taken from Table 2.
21
(8) The floods corresponding to 2-, 10-, 25-, 50-, 100-, 500-, and 1,000-year recurrence
intervals are computed with the three distribution methods described earlier, without any
modification of outliers and/or inliers, i.e., with the window as 0.
(9-11) The detection of outliers and inliers is initiated from window 1. Any detected
outlier and inliers are modified to correspond to the threshold level. The new Y series is
transformed to a y series then to a Q series with the previous value of λ. A new λ is derived and
the detection-modification process is repeated (usually two or three iterations) until no outliers
and inliers are detected in the window under consideration. The final Q or y series is used in
computing various T-year floods. The results are printed and the detection-modification process
is applied to the next window.
Examples
The methodology for flood frequency analysis with objective detection and modification
of outliers and inliers is applied to observed annual maximum floods for the Sangamon River
near Oakford (drainage area 5,093 square miles) using the 1941-1993 record. Tables 4 and 5
provide the computer-printout results, which are explained briefly.
Table 4 starts with U.S. Geological Survey (USGS) gaging station number, river name
and station location, and years of record, which determines the number of high or low floods
considered for detection of any outliers and inliers. The five highest observed floods were
123,000, 68,700, 55,900, 45,800, and 44,700 cfs, respectively. The highest flood of 123,000 cfs
is 1.79 times the next highest flood of 68,700 cfs and seems to be an outlier when the top five
floods are considered. The five lowest observed floods were 3,800, 5,670, 5,960, 8,400, and
10,000 cfs, respectively.
The " level number" refers to windows 0 - 6; 0 corresponds to no consideration of any
outliers/inliers, and windows 1 - 6 correspond to significance levels of 0.01, 0.05, 0.10, 0.20,
0.30, and 0.40, respectively. The objective detection and modification of any outliers and inliers
at various significance levels or windows is reflected in change of values of high and low floods.
The major modification in this example is in the highest flood (H1) of 123,000 cfs.
The distribution statistics are given for PT (power transformed) normal distribution, LP3
(log-Pearson HI) distribution, and MD (mixed) distribution. For the PT, skew is very close to
22
Table 4. Flood Frequency Analysis for Sangamon River near Oakford, Station Number 05583000, 1941-1993
Level No. Method 0 1 2 3 4 5 6
100-Year Flood in cfs Power Transform, PT
With kt = 3.0 85346 85346 82548 77773 73171 68979 66722 With sample kt 98095 98095 92916 84457 76900 70643 66704
LP III Sample skew 81399 81399 78539 73997 69810 66110 65482 Weighted skew 80065 80065 78581 75904 73191 70595 68506
MD 102869 102869 97879 88707 77775 72098 68827
Type Rank Observed and Modified Floods in cfs
Low 1* 3800 3800 3800 3800 3800 3824 4387 2* 5670 5670 5670 5670 5670 5714 6331 3* 5960 5960 5960 6239 6676 7045 7670 4* 8400 8400 8400 8400 8400 8400 8737 5* 10000 10000 10000 10000 10000 10000 9889
High 5* 44700 44700 44815 45607 46092 46092 46016 4* 45800 45800 47513 48385 48866 48866 48721 3* 55900 55900 55900 55900 55900 54856 52224 2* 68700 68700 68700 68700 66068 60647 57298 1* 123000 123000 112395 95194 81184 72095 66891
Method Statistics Values of Statistics
PT mean 21.053 21.053 25.717 37.177 56.491 89.013 93.108 stddev 2.351 2.351 3.209 5.506 9.791 17.725 18.197 skew 0.027 0.027 0.028 0.021 0.008 -0.003 -0.017 kurtosis 4.379 4.379 4.179 3.819 3.492 3.238 2.999 5th moment 2.393 2.393 1.984 1.176 0.504 0.128 -0.042 lambda 0.133 0.133 0.165 0.221 0.281 0.343 0.349
LP3 mean 4.348 4.348 4.348 4.347 4.346 4.345 4.346 stddev 0.271 0.271 0.270 0.266 0.261 0.257 0.247 skew -0.343 -0.343 -0.402 -0.489 -0.568 -0.638 -0.570 kurtosis 4.279 4.279 4.153 3.982 3.878 3.827 3.481 5th moment -2.479 -2.479 -3.359 -4.606 -5.624 -6.389 -5.246
MD weight'a' 0.510 0.510 0.522 0.519 0.407 0.219 0.204 mul 4.302 4.302 4.293 4.276 4.231 4.089 4.048 mu2 4.397 4.397 4.408 4.424 4.425 4.416 4.423 sigmal 0.345 0.345 0.337 0.324 0.317 0.285 0.231 sigma2 0.149 0.149 0.147 0.149 0.175 0.195 0.185 Test Stat 2.674 2.674 2.532 2.604 2.585 2.319 2.179
Notes: Drainage area of the Sangamon River near Oakford is 5,093 square miles. An asterisk indicates high and low floods considered for outlier detection and modification.
23
Table 5. Flood Frequency Analysis for Sangamon River near Oakford, Station Number 5583000, 1941-1993
Method Flood in cfs for Recurrence Intervals (years) Level 2 10 25 50 100 500 1000
PT,kt=3.0 0 22870 48599 62908 73981 85346 113061 125621 PT, sample kt 22870 46599 64156 79922 98095 150959 179053 LP3, sample skew 23110 48378 61652 71553 81399 104170 113959 LP3, weighted skew 23195 48228 61141 70671 80065 101482 110563 MD 23292 44920 62777 81142 102869 165145 197497
PT, kt=3.0 1 PT, sample kt LP3, sample skew same as above LP3, weighted skew MD
PT,kt=3.0 2 22976 48042 61629 72005 82548 107865 119179 PT, sample kt 22976 46371 62750 76959 92916 137393 160070 LP3, sample skew 23222 47863 60390 69560 78539 98796 107297 LP3, weighted skew 23219 47868 60407 69589 78581 98878 107401 MD 23479 44520 60964 77799 97879 155174 184754 PT,kt=3.0 3 23146 46954 59307 68543 77773 99406 108864 PT, sample kt 23146 45853 60136 71874 84457 117101 132736 LP3, sample skew 23376 46886 58229 66292 73997 90736 97507 LP3, weighted skew 23245 47126 58995 67580 75904 94407 102063 MD 23709 44060 57930 71736 88707 137921 163216 PT,kt=3.0 4 23298 45772 56940 65121 73171 91623 99531 PT, sample kt 23298 45138 57477 67042 76900 100858 111698 LP3, sample skew 23492 45849 56103 63190 69810 83683 89099 LP3, weighted skew 23248 46304 57504 65506 73191 90021 96886 MD 23584 44337 56151 66178 77775 113735 133606 PT,kt=3.0 5 23416 44577 54673 61932 68979 84813 91481 PT, sample kt 23416 44293 54936 62817 70643 88724 96526 LP3, sample skew 23558 44812 54114 60379 66110 77732 82125 LP3, weighted skew 23219 45452 56020 63488 70595 85946 92126 MD 23410 44142 55090 63495 72098 93320 103074 PT, kt=3.0 6 23418 43676 53242 60092 66722 81562 87790 PT, sample kt 23418 43686 53247 60093 66704 81537 87757 LP3, sample skew 23427 44046 53285 59611 65482 77681 82409 LP3, weighted skew 23195 44463 54554 61695 68506 83284 89266 MD 23556 43280 53399 61050 68827 87573 96032 Note: Drainage area of the Sangamon River near Oakford is 5,093 square miles.
24
zero (which is the purpose of power transformation) and kurtosis >3, indicating a sharper-peaked
frequency distribution than a normal distribution. All odd central moments are zero for a normal
distribution (Kite, 1977). In the present case, the fifth moment is generally positive, indicating
that the transformed series is not symmetrical. The transformation parameter (λ) value increases
with increase in window level. For the LP3, sample skew becomes progressively more negative
with increase in window level.
Much higher kurtosis than 3.0 indicates a sharper peak frequency curve than for the
normal distribution. For the MD, weight a remains close to 0.52 for windows 0 - 3 but drops off
for windows 4 and 5. The test statistic shows goodness of fit given by ∑|∆ Z|.
The 100-year flood peaks at various levels of significance or windows given for PT with
kt = 3.0 (assuming power-transformed series as a normal distribution) and with sample kt
(allowing for correction but considering power-transformed series as a symmetrical frequency
distribution), for LP3 with sample skew as well as weighted skew assuming a regional skew of
-0.4, and for MD. The 100-year flood values decrease as the level of detection or windows
increases for all the three distributions used.
Table 5 contains values of 2-, 10-, 25-, 50-, 100-, 500-, and 1,000-year flood peaks
derived with the three distributions at significance levels of 0.00, 0.01, 0.05, 0.10, 0.20,0.30, and
0.40, corresponding to windows 0, 1,2, 3, 4, 5, and 6, respectively. Because no outliers/inliers
were detected in window 1, the flood peak values in window 1 remain the same as in window 0.
Figure 7 shows the fitted distribution curves with MD (and two-component normal
distributions) and LP3 with sample skew as well as the observed peak floods for the 1941-1993
period for the Sangamon River near Oakford for window 3 or with outlier/inlier detection and
modification carried to 0.10 significance level. The LP3 does not satisfactorily fit the high end of
the flood spectrum.
The same methodology used for flood frequency was applied to stage frequency analysis.
As an example, the frequency method applied to annual maximum stages observed in the
Mississippi River at Grafton during 1941-1993 is considered here. Tables 6 and 7 provide the
computer print-out results.
Table 6 starts with USGS gaging station number, name of the river and station location,
and years of record. The five lowest stages (out of the 53 maximum annual stages) observed are
25
Figure 7. Sangamon River near Oakford, fitted flood probability curves
26
Table 6. Stage Frequency Analysis for Mississippi River at Grafton, Station Number 05585500,1941-1993
Level 0 1 2 3 4 5 6
Method 100-Year Stage in feet
PT With kt = 3.0 53.42 53.47 52.78 52.16 51.52 51.11 51.01 With sample kt 46.62 46.72 47.02 47.36 47.89 48.39 49.07
LP3 Sample skew 49.59 49.58 49.56 49.59 49.65 49.73 50.02 Weighted skew 46.72 46.73 46.75 46.82 46.96 47.14 47.47
MD 47.66 47.66 47.58 47.75 48.12 48.43 49.03
Type Rank Observed and Modified Stages, in feet
Low 1* 27.63 27.42 26.90 26.51 25.93 25.33 24.72 2* 27.69 27.69 27.52 27.20 26.73 26.25 25.76 3* 28.09 28.09 27.98 27.69 27.27 26.86 26.43 4* 28.09 28.09 28.09 28.09 27.71 27.33 26.95 5* 28.09 28.09 28.09 28.09 28.08 27.74 27.39
High 5* 40.88 40.88 40.88 40.88 40.88 40.96 41.45 4* 41.27 41.27 41.27 41.27 41.43 41.92 42.45 3* 41.55 41.55 41.55 41.95 42.63 43.15 43.76 2* 44.99 44.99 44.99 44.99 44.99 44.99 45.66 1* 49.90 49.90 49.90 49.90 49.90 49.90 49.90
Method Statistics Values of Statistics
PT mean 0.526 0.526 0.571 0.631 0.742 0.885 1.054 stddev 0.000 0.000 0.000 0.001 0.001 0.003 0.006 skew 0.079 0.074 0.071 0.072 0.061 0.045 0.033 kurtosis 1.964 1.971 2.031 2.102 2.221 2.345 2.482 5th moment 0.810 0.774 0.743 0.717 0.601 0.435 0.265 lambda -1.899 -1.897 -1.746 -1.579 -1.335 -1.106 -0.909
LP3 mean 1.525 1.525 1.525 1.525 1.525 1.524 1.524 stddev 0.063 0.063 0.064 0.064 0.065 0.067 0.068 skew 0.507 0.502 0.483 0.460 0.418 0.367 0.326 kurtosis 2.665 2.663 2.661 2.655 2.652 2.656 2.694 5th moment 4.307 4.272 4.128 3.926 3.542 3.080 2.651
MD weight'a' 0.345 0.345 0.345 0.275 0.265 0.264 0.204 mul 1.465 1.465 1.464 1.459 1.461 1.462 1.458 mu2 1.557 1.557 1.557 1.550 1.547 1.547 1.540 sigmal 0.019 0.019 0.016 0.012 0.018 0.021 0.024 sigma2 0.056 0.056 0.056 0.059 0.061 0.063 0.067 Test Stat 5.166 5.041 4.588 3.167 3.090 3.164 2.563
Notes: Drainage area of the Mississippi River at Grafton is 171,300 square miles. An asterisk indicates high and low stages considered for outlier detection and modification.
27
Table 7. Stage Frequency Analysis for Mississippi River at Grafton, Station Number 05585500,1941-1993
Method Level Stages in feet for Recurrence Intervals (years) 2 10 25 50 100 500 1000
PT,kt=3.0 0 32.91 40.68 45.20 49.03 53.42 67.46 76.59 FT, sample kt 32.91 41.31 43.88 45.37 46.62 48.98 49.85 LP3, sample skew 33.12 40.65 44.28 46.95 49.59 55.75 58.45 LP3, weighted skew 33.58 40.36 43.12 44.99 46.72 50.41 51.89 MD, mixed dist 32.93 41.16 44.01 45.91 47.66 51.37 52.88
•ft
PT,kt=3.0 1 32.90 40.69 45.22 49.06 53.47 67.57 76.76 PT, sample kt 32.90 41.31 43.91 45.43 46.72 49.15 50.06 LP3, sample skew 33.12 40.65 44.28 46.95 49.58 55.73 58.42 LP3, weighted skew 33.58 40.36 43.12 44.99 46.73 50.42 51.90 MD, mixed dist 32.92 41.17 44.02 45.91 47.66 51.37 52.88
PT,kt=3.0 2 32.93 40.69 45.09 48.72 52.78 64.92 72.10 PT, sample kt 32.93 41.25 43.96 45.60 47.02 49.81 50.87 LP3, sample skew 33.12 40.66 44.29 46.94 49.56 55.66 58.32 LP3, weighted skew 33.57 40.38 43.14 45.01 46.75 50.43 51.91 MD, mixed dist 32.90 41.13 43.96 45.84 47.58 51.27 52.77
PT,kt=3.0 3 32.95 40.72 44.98 48.42 52.16 62.73 68.52 PT, sample kt 32.95 41.21 44.03 45.80 47.36 50.50 51.72 LP3, sample skew 33.12 40.70 44.33 46.97 49.59 55.65 58.29 LP3, weighted skew 33.56 40.42 43.20 45.07 46.82 50.51 51.99 MD, mixed dist 33.19 41.09 44.00 45.95 47.75 51.60 53.17
PT,kt=3.0 4 32.98 40.80 44.91 48.13 51.52 60.50 65.03 PT, sample kt 32.98 41.20 44.20 46.14 47.89 51.55 53.02 LP3, sample skew 33.11 40.78 44.41 47.05 49.65 55.65 58.25 LP3, weighted skew 33.54 40.50 43.31 45.20 46.96 50.67 52.16 MD, mixed dist 33.02 41.16 44.19 46.23 48.12 52.18 53.83
PT, kt=3.0 5 33.01 40.91 44.90 47.96 51.11 59.03 62.82 PT, sample kt 33.01 41.24 44.38 46.46 48.39 52.51 54.18 LP3, sample skew 33.11 40.88 44.52 47.15 49.73 55.65 58.20 LP3, weighted skew 33.52 40.60 43.45 45.36 47.14 50.88 52.37 MD, mixed dist 32.91 41.25 44.37 46.47 48.43 52.61 54.33
PT, kt=3.0 6 33.04 41.09 45.04 48.01 51.01 58.34 61.72 PT, sample kt 33.04 41.34 44.68 46.94 49.07 53.69 55.61 LP3, sample skew 33.12 41.07 44.76 47.42 50.02 55.96 58.52 LP3, weighted skew 33.52 40.78 43.70 45.66 47.47 51.29 52.81 MD, mixed dist 33.03 41.43 44.72 46.95 49.03 53.53 55.37
Note: Drainage area of Mississippi River at Grafton is 171,300 square miles.
28
27.63, 27.69, 28.09, 28.09, and 28.09 feet above the gage datum. The five highest stages are
49.90, 44.99, 41.55, 41.27, and 40.88 feet. The highest stage of 49.90 feet is 4.91 feet higher
than the next highest stage, which in turn is 3.44 feet higher than the next highest stage.
However, the differences between the third and fourth highest and between the fourth and fifth
highest are only 0.28 and 0.39 feet, respectively.
The "level number" refers to windows 0 -6 , window 0 corresponds to no consideration of
any outliers/inliers, and windows 1-6 correspond to significance levels of 0.01, 0.05, 0.10, 0.20,
0.30, and 0.40, respectively. The objective detection and modification of any outliers/inliers at
various significance levels is reflected in the changes in values of low and high stages.
Table 6 provides the distribution statistics for PT, LP3, and MD. For PT, the skew is
close to zero and kurtosis is much lower than 3.0, indicating a flatter-peaked frequency
distribution than a normal distribution. The fifth moment is positive, indicating that the
transformed series is not symmetrical. The transformation parameter X algebraically increases
from -1.899 to -0.909 with the increase in window level. For the LP3, the sample skew
progressively decreases with increase in window level. Kurtosis values lie between 2.65 and
2.70, indicating flatter peaked frequency distribution than a normal distribution. For the MD, the
weight a remains at 0.409 from window 0 - 5 and reduces slightly in window 6. Test statistics
show a very good fit for windows 2 and 3.
The 100-year stage values at various levels of significance or windows are given for PT
with kt = 3.0 (assuming power-transformed series as a normal distribution) and with sample kt
(allowing for correction but considering power-transformed series as a symmetrical frequency
distribution), for LP3 with sample skew as well as weighted skew assuming a regional skew of
-0.4, and for MD. The 100-year stage values generally decrease with PT (kt = 3.0) but generally
increase with PT (sample kt), though significantly lower than the former. The 100-year stage
values with LP3 and MD do not change significantly from window to window though values
from LP3 (sample skew) are 2.87 to 2.55 feet higher than from LP3 (weighted skew) for
windows 0 - 6 because the sample skew varies from 0.507 to 0.326 but the regional skew is -0.4.
29
Table 7 contains values of 2-, 10-, 25-, 50-, 100-, 500-, and 1,000-year stages derived with the
three distributions at significance levels of 0.00, 0.01, 0.05, 0.10, 0.20, 0.30, and 0.40
corresponding to windows 0 - 6 , respectively.
Figure 8 shows the fitted distribution curves with MD (and two component normal
distributions), LP3 with sample skew, and observed annual peak stages for the 1941-1993 period
for the Mississippi River at Grafton for window 3 or with outlier/inlier detection and
modification carried to 0.10 significance level. The LP3 does not fit the observed peak stages
satisfactorily.
General Frequency Method
The power transformation is a useful tool in detection and modification of any outliers
and inliers. Various recurrence interval (T-year) floods and stages, however, cannot be estimated
accurately because of the problem with kurtosis values significantly different from 3.0 for the
normal distribution, and the asymmetrical frequency distribution as evidenced by fifth-order
moments not being close to zero. This makes kurtosis correction based on symmetrical
frequency distribution questionable.
The log-Pearson HI distribution with sample skew (when sample size is considerable, say,
more than 40 or 50) should be useful for analyzing the Illinois River floods and stages.
However, the generally used regional skew of -0.40, derived from natural stream flood records,
may not apply to managed rivers such as the Illinois River.
The mixed distribution fits the observed flood peaks and stages very well for most of the
gaging stations though at some stations the curves fitted by LP3 and MD are very close. For
flood and stage frequency analyses, both LP3 and MD results may be considered and evaluated
on their own merit.
The outlier/inlier detection and modification needs to be carried to an acceptable
significance level. One or two decades ago, a level of 0.01 was used. It can be interpreted in the
following words: if there are 100 sample of size n at one station, then one sample of these 100
samples will have a value as high or higher (or as low or lower) as the identified or perceived
outlier. This is a very drastic limitation and can lead to practical detection of no or very few
outliers/inliers. On the other hand, a significance level of 0.30 and 0.40 leads to another extreme.
30
Figure 8. Mississippi River at Grafton, fitted stage probability curves
31
It seems that a satisfactorily acceptable level can be taken as 0.10 or window 3. Thus, the results
presented hereafter from flood and stage analyses with the general frequency method are for
outlier/inlier detection and modification carried to the 0.10 significance level or window 3.
32
WATER STAGES AND ELEVATIONS IN ALTON POOL, ILLINOIS RIVER
There are six stage stations in the Alton Pool (River mile or RM 80.2 from LaGrange Lock
and Dam to RM 0.0 at the mouth of the Illinois River). Data for these stations together with their
U.S. Army Corps of Engineers (USCOE) numbers are given below.
USCOE Average bed Continuous Location number River mile level, ft-msl record
Dlinois River at Meredosia IM70 70.8 406.7 2/1884-12/1993 Illinois River at Valley City IVC61 61.3 406.6 1/1884-12/1993 Illinois River at Florence IF56 56.0 407.0 1/1942-12/1993 Dlinois River at Pearl IP43 43.2 408.0 1/1942-12/1993 Dlinois River at Hardin IH21 21.6 404.0 2/1932-12/1993 Mississippi River at Grafton 0218A -0.2 391.9 9/1929-12/1993
The maximum water surface elevations and their date of occurrence in the available
continuous record at each station from Meredosia to Grafton are: 446.69 feet above mean sea
level (feet-msl) on 5/26/43, 444.93 feet-msl on 5/26/43, 443.60 feet-msl on 8/1/93, 442.75 feet-
msl on 8/3-4/93, 442.30 feet-msl on 8/3/93, and 441.80 feet-msl on 8/1/93, respectively. It is
obvious that the maximum water elevations at Meredosia and Valley City occurred during the
highest flood at Meredosia, and those at Florence, Pearl, Hardin, and Grafton occurred during
unprecedented Mississippi River floods and stages in July-August 1993.
The USCOE at St. Louis provided daily water elevations (WE) for the period of record at
each station. These records give the annual maximum WE, number of days it persisted, and the
magnitude and date of the record maximum WE. Figures 9 and 10 plot the historical annual
maximum WE for Meredosia and Grafton.
Annual Maximum Water Elevations
After 1940, no new dams and levees were constructed along the Dlinois River
downstream of Peoria. Concurrent data at four stations (Meredosia, Valley City, Hardin, and
Grafton) cover calendar years 1941 - 1993, and data at two other stations (Florence and Pearl)
cover 1942 - 1993. Table 8 provides the annual maximum water elevations for all six stations.
33
Figure 9. Annual maximum water elevations, Illinois River at Meredosia
34
Figure 10. Annual maximum water elevations, Mississippi River at Grafton
35
Table 8. Annual Maximum Daily Water Elevations and Dates at Stage Stations in the Alton Pool, Illinois River (1941-1993)
Calendar Meredosia Valley City Florence Pearl Hardin Grafton year WE Date WE Date WE Date WE Date WE Date WE Date
1941 431.59(3) 11/11 430.00(3) 11/9 424.60(1) 11/9 423.09(1) 11/8
1942 437.40(1) 2/17 435.30(1) 2/17 433.70(2) 2/17 429.80(2) 2/17 426.10(1) 6/29 425.49(1) 6/29 424.50(1) 2/18 419.49(1) 2/17
1943 446.69(1) 5/26 444.91(1) 5/26 442.80(1) 5/26 439.10(1) 5/26 434.70(2) 5/24 432.78(1) 5/24
1944 443.19(2) 5/1 441.54(1) 5/2 440.02(1) 5/3 437.00(1) 5/1 433.90(1) 5/1 432.19(2) 4/30
1945 436.09(2) 5/23 434.30(1) 5/24 433.00(1) 5/23 430.60(2) 6/21 429.20(1) 6/13 427.79(1) 6/13 430.10(1) 5/23 427.40(2) 5/21 425.69(2) 5/20
1946 437.09(1) 1/15 435.21(1) 1/15 433.80(2) 1/14 431.00(1) 1/14 427.80(1) 1/14 425.49(1) 1/14
1947 438.79(1) 6/15 437.18(1) 6/15 436.20(1) 7/2 434.80(2) 7/2 433.40(2) 7/2 432.39(2) 7/2 437.79(1) 7/2 436.85(1) 7/2 436.20(1) 6/15
1948 438.79(1) 3/31 437.10(1) 3/29 435.90(1) 3/29 433.40(1) 3/28 430.80(2) 3/27 428.99(1) 3/27
1949 435.00(1) 2/27 432.90(1) 2/27 431.80(2) 2/7 429.30(4) 2/6 425.40(1) 2/2 420.19(1) 2/11 431.60(3) 2/27 425.30(4) 2/3 419.89(1) 2/4
1950 438.50(2) 5/3 436.50(1) 5/4 434.90(2) 5/3 431.30(2) 5/4 426.60(2) 5/3 422.79(1) 6/24 422.39 (3) 5/1
1951 438.09(2) 2/28 436.50(1) 7/24 435.50(1) 7/24 433.60(1) 7/24 431.80(1) 7/24 431.19(2) 7/20 437.50(1) 7/23
1952 436.00(2) 4/28 434.60(2) 4/29 433.60(2) 4/29 431.70(1) 4/30 429.80(2) 4/30 428.39(3) 4/30
Table 8. (Continued)
Calendar Meredosia Valley City Florence Pearl Hardin Grafton year WE Date WE Date WE Date WE Date WE Date WE Date
1953 429.90(2) 4/5 428.40(2) 4/5 427.50(1) 4/5 425.20(2) 4/5 422.50(1) 4/5 420.99(1) 4/5
1954 429.09(3) 4/24 427.50(4) 4/23 426.50(4) 4/23 424.00(5) 4/23 421.00(5) 4/23 419.99(2) 5/20 419.59(1) 4/22
1955 430.00(1) 4/29 428.50(1) 4/29 427.40(2) 4/29 424.80(4) 4/28 422.10(1) 4/25 420.19(1) 4/25
1956 427.50(3) 6/1 426.10(3) 5/31 425.20(2) 5/31 423.20(1) 5/31 421.00(1) 5/31 419.99(1) 4/30 419.39(1) 5/30
1957 436.10(2) 5/6 434.10(2) 5/6 432.70(2) 5/6 429.40(2) 5/7 424.80(3) 5/23 420.19(1) 6/15 424.50(3) 5/6 419.89(1) 5/25
1958 434.00(2) 6/23 432.10(1) 6/23 430.80(2) 7/23 428.10(2) 7/23 424.50(1) 8/3 421.09(2) 7/23 430.70(3) 6/23 427.70(2) 6/25 424.40(2) 7/23
1959 436.20(1) 2/20 434.30(1) 2/20 432.70(1) 2/20 429.50(1) 2/21 424.80(2) 2/21 420.69(1) 4/9 419.99(1) 2/21
1960 438.90(2) 4/8 437.20(2) 4/8 436.20(2) 4/10 434.00(1) 4/11 431.40(1) 4/10 429.49(1) 4/10
1961 433.70(1) 5/16 431.80(2) 5/15 430.60(2) 5/15 428.70(1) 5/11 426.30(1) 5/10 423.89(2) 5/10
1962 440.20(2) 3/29 438.30(2) 3/28 437.00(1) 3/29 433.80(2) 3/28 429.90(2) 3/26 426.99(1) 3/26
1963 431.40(1) 3/9 429.80(1) 3/9 428.80(1) 3/9 426.40(1) 3/6 422.80(2) 3/6 419.99(1) 3/7
1964 432.50(4) 4/28 430.70(3) 5/1 429.50(6) 4/27 426.90(5) 4/28 422.80(5) 4/28 419.59(2) 4/23 419.29 (1) 4/28
1965 435.70(1) 4/19 434.30(2) 4/19 433.50(2) 4/19 431.60(1) 4/19 429.50(1) 4/19 427.39(1) 4/19
Table 8. (Continued)
Calendar Meredosia Valley City Florence Pearl Hardin Grafton year WE Date WE Date WE Date WE Date WE Date WE Date
1966 436.00(1) 5/25 434.00(2) 5/25 432.70(2) 5/25 429.90(2) 5/21 426.10(2) 5/20 421.99(2) 5/20 429.90 (2) 5/25
1967 432.90(2) 12/29 431.20(2) 12/29 430.00(1) 5/10 427.90(2) 4/19 425.50(1) 4/19 423.69(1) 4/19 430.00(2) 12/28 427.50(4) 12/26 423.70(1) 12/24 419.59(1) 12/21
1968 436.40(1) 2/10 434.40(1) 2/12 433.00(3) 2/11 429.90(2) 2/11 426.40(1) 1/23 420.69(1) 1/24 427.60(2) 1/23 425.00(2) 2/12 419.09(1) 2/12
1969 435.50(2) 2/10 434.30(2) 7/13 433.70(1) 7/13 432.70(2) 7/13 431.50(3) 7/12 429.99(3) 7/12 435.00(2) 7/13
1970 440.50(1) 5/22 438.50(1) 5/22 437.10(1) 5/22 433.80(1) 5/2 429.50(1) 5/2 426.69(1) 9/28 433.80(2) 5/21 429.50(2) 5/18 426.39(1) 5/19
1971 436.05(1) 6/18 430.00(1) 3/2 429.30(1) 3/2 427.20(2) 3/3 424.50(1) 3/2 422.19(1) 3/4
1972 433.70(1) 5/1 431.90(2) 5/1 430.90(3) 4/30 428.40(2) 5/1 425.20(2) 5/5 422.59(1) 5/6
1973 444.70(1) 4/30 443.50(1) 4/30 442.40(1) 4/29 440.50(1) 4/29 438.20(1) 4/29 436.89(2) 4/28
1974 442.80(1) 6/30 440.80(1) 6/30 439.30(2) 6/30 436.10(1) 6/3 432.70(1) 6/3 429.56(1) 6/3 438.90(1) 6/2 426.58(1) 6/30
1975 435.10(2) 5/9 433.80(3) 5/9 432.90(3) 5/10 431.00(1) 5/10 428.50(1) 5/12 426.19(1) 5/12
1976 438.60(4) 3/12 436.70(1) 3/12 435.50(1) 3/16 431.90(3) 3/13 427.00(2) 4/30 425.76(1) 4/30 429.10(1) 5/2 419.18(1) 3/12
Table 8. (Continued)
Calendar Meredosia Valley City Florence Pearl Hardin Grafton year WE Date WE Date WE Date WE Date WE Date WE Date
1977 431.70(3) 5/7 431.00(1) 5/7 430.40(1) 5/7 428.50(1) 5/7 424.70(1) 5/8 421.28(1) 11/5 419.66(1) 5/7
1978 437.80(1) 5/20 436.00(1) 4/15 434.90(1) 4/15 432.00(1) 4/16 428.25(1) 4/16 424.94(2) 5/16 437.70(1) 4/15 436.00(2) 5/19 434.90(2) 5/19 431.80(3) 5/18 428.25(2) 5/16 424.70(1) 4/5
1979 445.10(3) 4/19 443.20(2) 4/19 441.90(1) 4/14 439.70(1) 4/14 436.50(1) 4/14 433.17(2) 4/14 444.30(1) 4/14
1980 435.70(3) 6/10 433.90(2) 6/11 432.90(2) 6/11 429.83(2) 6/12 425.00(4) 6/13 420.97(1) 6/19
1981 438.90(2) 5/23 436.90(2) 5/23 435.80(3) 5/22 433.08(1) 5/22 428.75(1) 5/21 425.29(1) 5/21
1982 444.30(1) 12/11 442.30(1) 12/12 441.00(1) 12/11 437.45(1) 12/11 433.75(2) 12/8 431.50(1) 12/8
1983 443.10(1) 4/14 441.40(1) 4/14 440.40(1) 4/14 437.55(1) 4/11 434.50(1) 4/10 432.05(1) 5/4 432.04 (1) 4/10
1984 439.70(1) 3/30 437.80(2) 3/29 436.70(2) 3/29 433.40(2) 3/29 430.00(1) 4/24 426.14(1) 4/27 428.50(4) 3/28 423.32(1) 3/29
1985 445.60(1) 3/10 443.60(1) 3/11 442.10(2) 3/10 438.55(1) 3/10 434.00(1) 3/9 430.47(1) 3/8
1986 439.30(1) 10/11 438.20(1) 10/11 437.70(1) 10/11 435.82(1) 10/10 434.50(2) 10/9 433.45(1) 10/9
1987 430.10(1) 12/31 428.70(2) 12/31 427.90(3) 12/31 425.40(1) 12/29 422.25(1) 12/29 420.72(1) 4/15 419.69 (1) 12/29
1988 431.80(1) 4/14 430.60(1) 1/15 430.20(1) 1/13 429.05(1) 1/10 425.25(3) 1/9 420.71 (1) 1/15 420.64 (1) 1/9
Table 8. (Concluded)
Calendar Meredosia Valley City Florence Pearl Hardin Grafton year WE Date WE Date WE Date WE Date WE Date WE Date
1989 429.00(1) 9/18 427.70(4) 9/16 427.00(3) 9/17 424.60(1) 9/17 421.15(1) 9/17 419.59(2) 3/21 419.22(1) 9/16
1990 438.30(1) 6/26 436.90(2) 6/26 436.00(2) 6/26 433.55(1) 6/26 430.65(1) 6/26 428.24(1) 6/26
1991 436.80(2) 5/7 434.90(2) 5/6 434.30(1) 5/7 431.80(1) 5/7 428.30(1) 5/7 425.23(1) 5/8
1992 434.50(1) 11/29 433.30(1) 11/29 432.40(2) 11/28 429.70(2) 10/27 426.40(2) 11/27 423.91(1) 12/19 423.87 (6) 11/26
1993 444.90(1) 7/28 444.10(1) 7/27 443.60(1) 8/1 442.75(2) 8/3 442.30(1) 8/3 441.80(1) 8/1
Notes: WE denotes maximum daily water elevation in a calendar year, in feet above mean sea level (ft-msl). A second line for a particular calender year indicates WE on a date closer to that for maximum WE at a preceding or succeeding stage station. The number in parentheses indicates the number of days that WE is maintained. For example 436.90 (2) 6/26 in the year 1990 for Valley City means that maximum daily stage of 436.90 ft-msl occurred for two days, beginning June 26, 1990.
The number in parentheses denotes the number of days the daily stage remained at the maximum
level. The date denotes the first day of the period (if more than one day) with maximum
elevation. A partial second line was added if the annual maxima did not occur near the same
date at all stations.
Figure 11 indicates annual maximum water elevations observed at Meredosia and Grafton
for the period 1941-1993 for the months of January through December. About 85% of observed
maxima at Meredosia occurred from February to June, and about 79% occurred at Grafton from
March to July. Of the highest ten values observed at Meredosia, eight occurred from March to
June, the rank 4 value occurred in July 1993 because of exceptionally high backwater from the
Mississippi River, and the rank 6 value occurred in December 1982. None of these ten values
occurred during 1945 to 1969. At Grafton, seven of the ten highest WE values occurred from
April to July, the highest observed WE so far occurred on August 1, 1993, and was 4.91 feet
higher than the second highest on April 28, 1973, the rank 3 value occurred on October 9, 1986,
and the rank 9 value occurred on December 8, 1982. None of these ten values occurred during
1952 to 1972.
Meredosia Grafton
Rank WE,ft-msl Date WE,ft-msl Date
1 446.69 5/26/43 441.80 8/1/93 2 445.60 3/10/85 436.89 4/28/73
3 445.10 4/19/79 433.45 10/9/86
4 444.90 7/28/93 433.17 4/14/79
5 444.70 4/30/73 432.78 5/24/43
6 444.30 12/11/82 432.39 7/2/47
7 443.19 5/1/44 432.19 4/30/44
8 443.10 4/14/83 432.05 5/4/83
9 442.80 6/30/74 431.50 12/8/82
10 440.50 5/22/70 431.19 7/24/51
WE Rankings for Stations in Alton Pool. Table 9 provides the year and date of ranked
annual maximum water elevations (in descending order of magnitude, rank 1 being the highest)
and ranks of associated flood peaks at Meredosia. Also included are the ranks of maximum
41
Figure 11. Annual maximum water elevations at Meredosia and Grafton, 1941-1993
42
Table 9. Ranked Annual Maximum Water Elevations and Floods at Meredosia and Related Water Elevation Ranks at Other Station in Alton Pool, Dlinois River
Meredosia WE Meredosia Flood WE Ranks at Rank Date Rank Date Valley City Florence Pearl Hardin Grafton ∆WE
43
13.91
15.13
11.93
3.10
7.81
12.80
11.00
11.06
16.22
14.11
13.21
16.38
5.85
9.41
13.61
9.80
5.40
19.42
16.11
10.06
5
11
4
1
2
9
7
6
(21)
(22)
20
(33)
3
14
27
15
6
(52)
(34)
17
4
7
3
1
2
9
8
5
11
10
17
16
6
14
23
15
10
28
27
22
4
5
3
1
2
7
8
6
9
13
14
17
10
12
19
18
11
22
26
16
2
4
5
1
3
6
8
7
9
11
12
13
10
14
18
17
15
19
21
16
1
3
5
2
4
6
7
8
9
10
11
13
12
14
17
16
15
19
20
18
1
2
4
13
7
3
8
(11)
5
10
11
17
24
22
18
21
25
12
15
19
123,000
122,000
111,000
+80,000
102,000
112,000
102,000
94,000*
110,000
94,800
91,900
76,800
69,600
73,000
75,500
73,600
68,300
80,600
77,200
74,460
5/26
3/10
4/19
7/28
4/30
12/11
5/1
4/14
6/30
5/22
3/29
3/30
10/11
4/8
5/23
3/31
6/15
3/12
5/3
6/26
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
1943
1985
1979
1993
1973
1982
1944
1983
1974
1970
1962
1984
1986
1960
1981
1948
1947
1976
1950
1990
Notes: AWE = difference in water elevations at Meredosia and Grafton in feet. * = average daily flow corresponding to maximum stage at Meredosia.
() = rank for the relevant flood and/or stage if annual maximum occurs in a different period than at Meredosia.
+ = estimated flood peak at Meredosia.
water elevations occurring at Valley City, Florence, Pearl, Hardin, and Grafton for the year and
related event at Meredosia. In 1983, mean daily flow (94,000 cfs) corresponding to maximum
stage at Meredosia had a relative ranking of 11, though the peak discharge (104,000 cfs) on
March 24 had a rank of 6. At Grafton, for 5 of the 20 years listed, the peak stage did not occur
around that at Meredosia. Relative ranks for stages occurring around the Meredosia dates are
given in parentheses. The differences between water elevations at Meredosia and Grafton, or
AWE in feet (feet), are given in the last column of the table. The following inferences can be
made.
1. When the rank number decreases from Meredosia to Grafton, the water elevation at
Grafton considerably raises elevations upstream though the effect becomes attenuated
with distance. This situation, when combined with relatively low flood peak at
Meredosia but very high water elevation at Grafton, causes serious backwater effects
from the Mississippi River, extending upstream to Meredosia. For example, in 1993 the
Meredosia flood peak rank was 13 or about a 5-year flood, but the Grafton stage peak
rank was 1 (higher by 4.91 feet over the rank 2 event) and caused the highest water
elevations at Hardin, Pearl, and Florence, second highest at Valley City, and fourth
highest at Meredosia. The AWE was only 3.1 feet, the minimum in the 53-year record,
instead of 10 to 16 feet during moderate and low backwater effects. Similar, but less
dire conditions occurred in 1947, 1986, and 1973. Had the maximum flood (123,000
cfs) occurred in 1993, water elevations at Meredosia and some stations downstream
would have increased by about 3 feet or more, overtopping all six levees below
Meredosia and probably some levees upstream.
2. When the rank number increases from Meredosia to Grafton, the backwater effects from
the Mississippi River are moderate to low, depending on the relative increase in rank
number. The maximum value of AWE in 20 years (Table 9) was 19.42 feet when the
Grafton water elevation approached the lowest annual maximum.
3. Monthly distribution of maximum water elevations at Meredosia for the top 20 events is:
5 in March, 4 in April, 5 in May, 3 in June, and one each in July, October, and
December. Two of the top six water elevations at Meredosia had very significant
44
backwater effects from the Mississippi River. The top three ranked floods occurred
from March though May.
4. Ranks at Meredosia and Valley City are very similar, but this does not hold true for
ranks at Florence, Pearl, Hardin, and Grafton where the Mississippi backwater effects
vary in severity and upstream-reach length zone of influence.
Stage Frequency Analysis
The annual maximum stage series were derived from the annual maximum water
elevation or WE series by subtracting the average bed level (ABL) from the water elevation at the
stage station. These series at Meredosia, Valley City, Florence, Pearl, Hardin, and Grafton were
run on the general frequency program. Table 10 provides the parameter values of fitted mixed
distribution and log-Pearson III distribution with sample skew. The values of kurtosis for the
corresponding power-transformed series are also included. These values are much lower than 3.0
for the normal distribution, indicating that the top portion of the frequency distribution curve for
the power-transformed series (with practically zero skew) is flatter than for a normal distribution.
The following low (L) and high (H) stages were slightly modified in the third window (0.1
significance level) at each station.
Modified Station HorL from to
Meredosia H5 38.00 37.74 Valley City H5 36.60 36.20 Florence H5 34.90 34.49 Pearl None Hardin None Grafton H3 41.55 41.95
L3 28.09 27.69 L2 27.69 27.20 L1 27.63 26.51
H5 indicates fifth highest value, L2 indicates second lowest value, and so on.
45
Table 10. Parameters for Fitted Mixed and Log-Pearson III Distributions
Stage station
Parameter Meredosia Valley City Florence Pearl Hardin Grafton
Mixed distribution a 0.872 0.925 0.415 0.831 0.913 0.409
µ1 1.488 1.458 1.387 1.359 1.382 1.506
σ1 0.059 0.066 0.066 0.086 0.083 0.051
µ2 1.374 1.336 1.454 1.386 1.320 1.540
σ2 0.039 0.034 0.066 0.063 0.005 0.068
Log-Pearson III distribution
µ 1.474 1.449 1.426 1.364 1.377 1.525
σ 0.068 0.071 0.073 0.083 0.082 0.064
g -0.159 -0.070 -0.112 -0.081 0.218 0.460
Power-transformed series
Kurtosis 2.679 2.625 2.673 2.813 2.591 2.102
Notes: 1. Distribution parameters apply to log-transformed stages in the third window of outlier/inlier modification. 2. Modifications at Grafton included L1, L2, and L3 (from 27.63, 27.69, and 28.09 feet to 26.51, 27.20 and
27.69 feet, respectively), and H3 (from 41.55 to 41.95). Modifications at Florence, Valley City, and Meredosia included H5 (from 34.90, 36.60 and 38.00 feet to 34.49, 36.20 and 37.74 feet, respectively).
3. Power transformation modifies stage series to be close to normal, with skew very close to zero. A normal series has kurtosis equal to 3.0. A lesser kurtosis means the modified series are platykurtic or have flatter tops than the normal frequency distribution.
46
Figures 12-17 plot the fitted mixed distribution (and its two-component normal
distributions), fitted log-Pearson HI distribution, and observed peak stages for the Meredosia,
Valley City, Florence, Pearl, Hardin, and Grafton stations. Both distributions fit observed stages
at Meredosia, Valley City, Florence, and Pearl equally well, but the fit with the mixed
distribution is better than with the log-Pearson III distribution for the observed stages at Hardin
and Grafton.
Table 11 provides various recurrence-interval or T-year maximum stages with both mixed
and log-Pearson IE distribution at each of the stations. It also includes the two highest stages
observed during the period 1941/1942 to 1993. Values of T-year stages up to 100-year
recurrence interval from both the distributions at Meredosia, Valley City, Florence, and Pearl are
very similar (differences within a range of -0.38 to 0.32). The range is a little higher for Hardin
and Grafton in the case of T=50 and T=100 years. However, values from mixed and log-
Pearson HI distributions for T=500 and 1,000 years are significantly different at Hardin and
Grafton. Values from mixed distribution provide a better fit and seem to be more realistic and
viable in terms of the historic data. Relevant T-year stage values were used to determine T-year
water elevations at all six stations.
Table 12 provides the highest two stages, S1 and S2, and their date of occurrence at
Grafton, Hardin, Pearl, Florence, Valley City, and Meredosia. The recurrence intervals T1 and
T2 were fit to S1 and S2 using mixed distribution parameters and are also included in the table.
The highest stages from Grafton to Meredosia are assigned recurrence intervals of 193, 126, 64,
33, 38, and 43 years in a record of 52 to 53 years. It is interesting to note that the fitted
recurrence interval T is the highest at Grafton (193 years), followed by Hardin (136 years) and
Pearl (64 years). The 1993 unprecedented high stage, 4.91 feet higher than the next highest stage
at Grafton, greatly increased T at Hardin to 136 years and at Pearl to 64 years. The 1993 flood
produced a stage at Grafton much higher than expected during a 53-year record. The backwater
effect from the Mississippi River was so strong that it greatly increased T1 at Hardin and
moderately increased T1 at Pearl, an effect that attenuates with distance upstream.
47
Figure 12. Fitted mixed and log-Pearson HI distributions to observed annual maximum stages, Illinois River at Meredosia, 1941-1993
48
Figure 13. Fitted mixed and log-Pearson HI distributions to observed annual maximum stages, Illinois River at Valley City, 1941 -1993
49
Figure 14. Fitted mixed and log-Pearson HI distributions to observed annual maximum stages, Illinois River at Florence, 1942-1993
50
Figure 15. Fitted mixed and log-Pearson HI distributions to observed annual maximum stages, Illinois River at Pearl, 1942-1993
51
Figure 16. Fitted mixed and log-Pearson III distributions to observed annual maximum stages, Illinois River at Hardin, 1941-1993
52
Figure 17. Fitted mixed and log-Pearson HI distributions to observed annual maximum stages, Mississippi River at Grafton, 1941-1993
53
Table 11. Various Recurrence - Interval Stages in Feet at Stage Recording Stations in Alton Pool, Illinois River
Stages at recurrence interval, ft-msl Two highest
Distribution 2-year 10-year 25-year 50-year 100-year 500-year 1000-year stages recorded
Mississippi River at Grafton (1941-1993)
MD 33.33 40.62 43.91 46.19 48.34 52.96 54.83 49.90(1993)
LP3s 33.12 40.70 44.33 46.97 49.59 55.65 58.29 44.99(1973)
Illinois River At Hardin (1941-1993)
MD 23.58 30.55 33.49 35.53 37.47 41.71 43.47 38.30(1993)
LP3s 23.66 30.48 33.64 35.91 38.12 43.18 45.36 34.20(1973)
Illinois River at Pearl (1942-1993)
MD 23.18 29.44 32.15 34.07 35.91 40.01 41.72 34.75(1993)
LP3s 23.17 29.45 32.08 33.89 35.59 39.24 40.73 32.50(1973)
Illinois River at Florence (1942-1993)
MD 26.73 33.19 35.85 37.66 39.35 42.97 44.44 36.60(1993)
LP3s 26.77 33.06 35.62 37.36 38.97 42.39 43.76 35.80(1943)
Illinois River at Valley City (1941-1993)
MD 28.25 34.62 37.24 39.01 40.69 44.28 45.74 38.31 (1943)
LP3s 28.14 34.63 37.31 39.13 40.83 44.46 45.93 37.50(1993)
Illinois River at Meredosia (1941-1993)
MD 30.02 36.22 38.68 40.34 41.89 45.19 46.53 39.99(1943)
LP3s 29.88 36.28 38.83 40.53 42.09 45.37 46.57 38.90(1985)
Notes: MD denotes values from stage frequency analysis with mixed distribution. LP3s denotes values from stage frequency annalysis by Log-Pearson HI distribution with sample skew.
54
Table 12. Observed Maximum Stages, Dates and Fitted Recurrance Intervals for Stations in Alton Pool, Illinois River
Continuous Av. bed historical 1993 peak
level S1 S2 T1 T2 record used stage and Stage Station RM (ft-msl) (ft) (ft) (yrs) (yrs) rank
Mississippi River at Grafton 0.0 391.90 49.90 44.99 193 31 1941-1993 49.90: 1 (8/1/93) (4/28/73)
Illinois River at Hardin 21.6 404.00 38.30 34.20 136 32 1941-1993 38.30: 1 (8/3/93) (4/29/73)
Illinois River at Pearl 43.2 408.00 34.75 32.50 64 28 1942-1993 34.75: 1 (8/1/93) (4/29/73)
Illinois River at Florence 56.0 407.00 36.60 35.80 33 25 1942-1993 36.60: 1 (8/1/93) (5/26/43)
Illinois River at Valley City 61.3 406.60 38.31 37.50 38 28 1941-1993 37.50: 2 (5/26/43) (7/27/93)
Illinois River at Meredosia 70.8 406.71 39.99 38.90 43 27 1941-1993 38.20: 4 (5/26/43) (3/10/85)
Table 13. Various Recurrence - Interval River Water Elevations with MD at Stage Stations in Alton Pool, Dlinois River
Water surface elevations, ft-msl, at recurrance interval Max. Observed
2-year 10-year 25-year 50-year 100-year 500-year 1000-year elevation & year
Mississippi River at Grafton. RM = 0. and bed level = 391.9 ft-msl
425.23 432.52 435.81 438.09 440.24 444.86 446.73 441.80(1993)
Illinois River at Hardin. RM = 21.6. and bed level = 404.0 ft-msl
427.58 434.55 437.49 439.53 441.47 445.71 447.47 442.30(1993)
Illinois River at Pearl. RM = 43.19. and bed level = 408.0 ft-msl
431.18 437.44 440.15 442.07 443.91 448.01 449.72 442.72(1993)
Illinois River at Florence. RM = 56.0. and bed level = 407.0 ft-msl
433.73 440.19 442.85 444.66 446.35 449.97 451.44 443.60(1993)
Illinois River at Valley City. RM = 61.3. and bed level = 406.6 ft-msl
434.85 441.22 443.84 445.61 447.29 450.88 452.34 444.91 (1943)
Illinois River at Meredosia. RM = 70.8 and bed level = 406.7 ft-msl
436.72 442.92 445.38 447.08 448.59 451.89 453.23 446.69(1943)
55
T- Year River Water Elevations
Maximum water surface elevations for T equal to 2 to 1,000 years at each station were
developed by adding corresponding stages to average river bed level. Table 13 provides this
information and includes river mile and average bed elevation at each station, and maximum
observed elevation and the year in which it occurred. Figure 18 provides the T-year elevations
and the curve joining the points for a particular T defines the profile of the T-year flood
elevations. The difference in water surface elevation from Meredosia to Grafton, or AWE, is
given for each profile.
T, years 2-year 10-year 25-year 50-year 100-year 500-year 1,000-year
AWE, ft 11.49 10.40 9.57 8.99 8.35 7.03 6.50
Flood water profiles in Figure 18 apply if the same T-year stage or water elevation occurs
around the same date at all stations. This is not always true because of the absence of a linear
relationship between maximum water elevation at Grafton and maximum flood at Meredosia.
The figure provides water profiles for the 1993 and 1943 flood events (1993 with maximum
stage at Grafton and 1943 with maximum flood at Meredosia) for comparison with other T-year
profiles. Maximum water surface elevations at four downstream stations (Florence, Pearl,
Hardin, and Grafton) occurred in the beginning of August 1993 during the unprecedented
Mississippi River flood and its backwaters upstream along the Illinois River. Maximum water
elevations occurred at Valley City and Meredosia during the last week of May 1943.
Grafton Stages and Meredosia Flood Peaks
Water elevation or stage at Grafton and corresponding flood peak at Meredosia govern
the flood profile in the Illinois River Alton Pool. Figure 19 provides Meredosia flood peak ranks
versus Grafton maximum water elevation ranks, and curve A gives the worst historical joint
probability of heavy flooding in the Alton Pool. Considering 100 years as the maximum design
event, the interest lies in 25-,50-, and 100-year peak floods and stages for simulating the flood
profiles using the UNET model (Barkau, 1993).
56
Figure 18. T-year maximum water elevation profiles for Illinois River from Meredosia to Grafton (T is average recurrence interval in years)
57
Figure 19. Meredosia flood peak ranks versus Grafton maximum water elevation ranks, using 1941-1993 data
58
Peak Flood at Meredosia Corresponding Maximum Water Elevation, Grafton
Rank T, years Rank T, years
1 100 5.0 20
2 50 4.6 22
3 33 4.2 24
4 25 3.8 26
Thus, for maximum flood peaks at Meredosia corresponding to T of 100, 50, and 25 years, the
design T for maximum elevations at Grafton may be taken as 25 years. The following
combinations of Tf (recurrence interval for flood peaks at Meredosia) and Tw (associated
recurrence interval for maximum flood elevations at Grafton) provide various scenarios.
(T f , Tw): (100, 25), (100, 50), and (100, 10)
(T f , Tw): (50, 25), (50,50), and (50, 10)
(T f , Tw): (25,25), (25, 50), and (25, 10)
For maximum water elevations at Grafton, the following joint probabilities of Tw, Tf may
be considered.
(Tw , Tf): (100, 10), (100, 25), (100, 50)
(Tw , Tf): (50, 25), (50, 10)
Various combinations may be simulated to analyze chances of levee damage and failure,
and to consider suitable, efficient, and effective measures to enhance safety of structures and to
minimize loss of life and property in case of a levee break. These combinations include:
(T f , Tw): (100, 50), (100, 25), (100, 10)
(T f , Tw): (50,100), (50, 50), (50, 25), (50, 10)
(T f , Tw): (25,100), (25, 50), (25, 25), (25, 10)
(T f,Tw): (10, 50), (10, 100)
These simulated flood profiles also need to be validated for the 1993 and 1943 flood events.
59
Highest Observed Water Surface Profiles, Alton Pool, Illinois River
Figure 20 provides the observed water surface profiles (joining maximum water
elevations at six stage stations in the Alton Pool) for the eight highest water elevations at
Meredosia in 1943, 1985, 1979, 1993, 1973, 1982, 1944, and 1983. Each profile also includes
the Meredosia rank and associated flood peak rank as well as the Grafton rank. The difference in
maximum water elevations (AWE) at Meredosia and Grafton are also given for each event.
These eight profiles cover Grafton WE ranks of 5, 11,4, 1, 2, 9, 7, and 8, respectively. Grafton
WE ranks of 3 and 6 are not included. Thus, two more profiles with the missing ranks at Grafton
are also included. The information presented in Figure 20 is given in a tabular form below.
WE Rank
Year Meredosia Grafton Meredosia Qp Rank AWE, ft
1943 1 5 1 13.91
1985 2 11 2 15.13
1979 3 4 4 11.93
1993 4 1 13 3.10
1973 5 2 7 7.81
1982 6 9 3 12.80
1944 7 7 8 11.00
1983 8 8 11 11.06
1986 13 3 24 5.85
1947 17 6 25 5.40
Note: Qp denotes the flood peak at Meredosia.
The lower the AWE, the more pronounced are the Mississippi River backwater effects. In
1993, maximum backwater effects were felt up to Meredosia from June to September during
unprecedented high water levels and floods in the Mississippi River. A minimal backwater
effect was felt in the 1985 event (out of the ten listed) when the rank of Grafton WE was 11.
60
Figure 20. Highest observed water surface profiles, Meredosia to Grafton, 1941-1993 (WE3 is WE with rank 3, Q3 is flood peak with rank 3, etc.)
61
Observed Stage Hydrographs for 1943 and 1993 Events
The 1943 event was marked by the historical maximum flood (123,000 cfs), on May 26-
28 and stage or water elevation WE, though the corresponding WE at Grafton ranked fifth in the
1941-1993 record. The dates of maximum WE and their ranks from Meredosia to Grafton are as
follows.
Location Max WE, ft-msl Rank Date
Illinois River at Meredosia 446.69 1 5/26/43 Illinois River at Valley City 444.91 1 5/26/43
Illinois River at Florence 442.80 2 5/26/43
Illinois River at Pearl 439.10 4 5/26/43
Illinois River at Hardin 434.70 4 5/24/43
Mississippi River at Grafton 432.78 5 5/24/43
Note: Maximum WE at Hardin lasted two days (May 24 and 25).
Figure 21 provides daily water elevations at the six stations from May 17 to May 31,
1943 which were used to draw WE hydrographs. The daily flows at Meredosia are also plotted
and connected to yield the flood hydrograph. The figure shows that:
1. The water elevation at Grafton reached its maximum value on May 24 and started
declining thereafter, at a steep rate from May 26 to May 31. The effect of this decline
was an increase in water surface slope and a reduction in water surface elevations at
Hardin and Pearl.
2. The flood discharge remained about 123,000 cfs at Meredosia from May 26 to May 28.
The corresponding daily water elevations at Meredosia were 446.69, 446.59, and 446.40
feet-msl, respectively. The steep decline of 0.4, 0.5, and 0.5 feet per day occurred during
the next three days.
3. Because of limited backwater effects from the Mississippi River, the water elevation
values at Valley City, Florence, Pearl, Hardin, and Grafton show a steadily increasing
rate of decline over time.
4. The difference in maximum water elevation at Meredosia and Grafton, 13.91 feet,
substantiates that Mississippi backwater effects were not considerable. The rank of
62
Figure 21. Water elevation versus time curves at Meredosia, Valley City, Florence, Pearl, Hardin, and Grafton during May 17-31, 1943
63
maximum WE was 5 at Grafton and 1 at Meredosia. The ranks in Alton Pool steadily
dropped from 1 at Meredosia/Valley City to 5 at Grafton.
The 1993 event, on the other hand, presents a classic example of the tremendous
backwater effects from the Mississippi River. Unprecedented floods in the Mississippi River in
July and August caused maximum water elevations of record at Grafton, Hardin, Pearl, and
Florence. The dates of maximum water elevations and their ranks from Meredosia to Grafton are
given below.
Location Max WE, ft msl Rank Date
Illinois River at Meredosia 444.91 4 7/28/93 Illinois River at Valley City 444.10 2 7/27/93
Illinois River at Florence 443.60 1 8/1/93
Illinois River at Pearl 442.75 1 8/3/93
Illinois River at Hardin 442.30 1 8/3/93
Mississippi River at Grafton 441.80 1 8/1/93
Note: Maximum water elevation at Pearl lasted two days (August 3 and 4).
Figure 22 provides daily water elevations at the six stations from July 25 to August 8,
1993, which were used to draw WE hydrographs. The daily flows at Valley City (Meredosia was
dropped as a discharge station in October 1989) are also plotted and connected to yield the daily
flood hydrograph. The observed peak on August 1 was 92,000 cfs as a result of levee breaks, and
the flow was accelerated because of a drop in water elevation at the breaks and consequent
steepening of the water surface slope. Without the levee break the flood peak would have been
about 82,000 cfs at Valley City and about 80,000 cfs at Meredosia. This peak flow will have a
rank of 13 among the annual maximum floods at Meredosia from 1941-1993. The stage at
Valley City was affected by the Mississippi backwaters from June to September 1993. Figure 22
shows that:
1. The water elevation graphs are much flatter than those for the 1943 event (Figure 21), a
result of exceptionally high and prolonged WEs at Grafton, which experienced a rank of
2 (436.89 feet-msl), in 1973 that was 4.92 feet lower than that in 1993.
64
Figure 22. Water elevation versus time curves at Meredosia, Valley City, Florence, Pearl, Hardin, and Grafton during July 25 to August 8, 1993
65
2. The WE hydrographs show a dip on August 1 at Pearl, and on August 2 at Florence,
Valley City, and Meredosia. These dips were caused by failure of the Eldred, Hartwell,
and Hillview levees (approximately from RM 24 to RM 50) on August 1 after
overtopping due to backwaters from the Mississippi River. Waters pouring into areas
protected by these levees lowered the WE at Pearl on August 1, and the effect reached
upper stations the next day.
3. The difference in maximum water elevations at Meredosia and Grafton is only 3.10 feet,
the lowest amount in the 53-year record. The rank of maximum water elevation at
Grafton was 1, which was maintained by backwaters up to Florence, and then it dropped
to 3 at Valley City, and 4 at Meredosia. This is the reverse of what happened in 1943.
66
FLOOD FREQUENCY ANALYSES FOR MAJOR TRIBUTARIES
. Five major tributaries discharge into the Illinois River downstream of Peoria: Mackinaw
River near Green Valley, Spoon River at Seville, Sangamon River near Oakford, La Moine River
at Ripley, and Macoupin Creek near Kane. For each of these tributaries, Table 14 provides the
U.S. Geological Survey (USGS) gaging station number, period for which the continuous
historical flood record is available, size of drainage area (DA), station location and distance
upstream of mouth or confluence with the Illinois River, drainage area at the mouth, and Illinois
River mile (RM) at the confluence.
The general frequency program was used to develop distribution parameter values (mixed
and log-Pearson HI distributions) for the entire record as well as split samples to examine the
trends, if any, in various frequency floods for each of the five tributaries. The results are given
for the parameter values (Table 15) and for 2-, 10-, 25-, 50-, 100-, 500-, and 1,000-year flood
peaks fitted with mixed distribution as well as log-Pearson HI (LP3) distribution (Table 16). Any
outliers/inliers detected at the upper and lower end of the flood spectrum are modified at the 0.1
significance level (Table 17).
Most of the modifications are minor except for the highest flood of 123,000 cubic feet per
second (cfs) that occurred in the Sangamon River near Oakford on May 20, 1943. The second
and third highest floods were 68,700 cfs on December 12, 1982 and 55,900 cfs on April 15,
1979, respectively. Corresponding flood peaks at Meredosia were 123,000, 112,000, and
111,000 cfs with ranks of 1, 3, and 4, respectively. The second highest flood (122,000 cfs at
Meredosia) occurred on March 10, 1985; a corresponding flood peak (78,800 cfs at Kingston
Mines at RM 144.4) occurred on March 6, 1985. Flood peaks in the Spoon, Sangamon, and La
Moine Rivers (between Kingston Mines and Meredosia) at their respective gaging stations were
29,200 cfs on March 6, 26,500 cfs on February 25, and 28,000 cfs on March 7, 1985.
Design T-Year Floods
Figures 23-27 show fitted mixed distribution (together with two-component normal
distributions) and log-Pearson III with sample skew (LP3s) distribution to the observed annual
floods for the period 1941-1993 for the five USGS gaging stations. Distributions for the
67
Table 14. Relevant Data for Major Tributaries to Illinois River below Peoria, Illinois
Upstream DA at USGS Period of DA, of mouth, IL River
Tributary station no. record sq mi mouth, mi sq mi mile
Mackinaw River 05568000 1922-1993 1089 13.7 1136 147.7 near Green Valley
Spoon River 05570000 1919-1993 1636 38.7 1855 120.4 at Seville
Sangamon River 05583000 1931-1993 5093 25.7 5419 88.9 near Oakford
La Moine River 05585000 1921-1993 1293 12.3 1350 83.5 at Ripley
Macoupin Creek 05587000 1941-1993 868 16.1 961 23.2 near Kane
68
Table 15. Fitted Mixed Distribution (MD) and Log-Pearson III (LP3s) Distribution Parameters
River and MD parameters LP3s parameters
station Period a µ1 σ1 µ2 σ2 µ a g Mackinaw River near Green Valley
1922-1993 0.283 3.704 0.120 4.023 0.307 3.932 0.304 0.450
1922-1951 0.545 3.640 0.145 4.213 0.183 3.901 0.329 0.265
1941-1993 0.285 3.776 0.105 4.048 0.315 3.970 0.299 0.486
1973-1993 0.101 3.909 0.525 4.119 0.303 4.097 0.338 -0.289 Spoon River at Seville
1919-1993 0.445 3.907 0.176 4.232 0.168 4.088 0.236 -0.122
1941-1993 0.308 3.954 0.195 4.192 0,179 4.119 0.214 -0.206 1973-1993 0.272 3.968 0.272 4.225 0.199 4.155 0.249 -0.444
Sangamon River near Oakford
1931-1993 0.548 4.194 0.333 4.463 0.142 4.315 0.296 -0.675 1941-1993 0.519 4.276 0.324 4.424 0.149 4.347 0.266 -0.489
1973-1993 0.532 4.449 0.230 4.386 0.148 4.419 0.198 0.188
La Moine River at Ripley
1921-1993 0.286 3.736 0.285 4.0.37 0.209 3.951 0.270 -0.474
1941-1993 0.499 3.869 0.261 4.140 0.167 4.005 0.258 -0.480
1973-1993 0.169 3.651 0.334 4.167 0.182 4.079 0.290 -1.224 Macoupin Creek near Kane
1941-1993 0.549 3.778 0.369 4.165 0.170 3.952 0.354 -0.666
1973-1993 0.124 3.550 0.149 4.094 0.194 4.027 0.261 -0.589
Notes: MD = mixed distribution (µ1 and σ1 are mean and standard deviation of one component distribution, µ2 and σ2 of second component distribution, a is weight of first component distribution and 1-a of second component), LP3s = Log-Pearson Type III distribution with sample skew (µ, σ, and g denote mean, standard deviation, and skew)
Table 16. Flood Frequency Analysis for Major Tributaries to Illinois River below Peoria, Illinois
Observed Floods at various recurrence intervals, cfs Record Qmax Method 2-year 10-year 25-year 50-year 100-year 500-year 1000-year
Mackinaw River near Green Valley, 1075 sq mi
1922-1993 51,000 MD 7,700 22,674 32,484 40,788 49,953 74,928 87,992
LP3s 8,122 21,571 32,274 42,433 54,770 94,303 117,327
LP3w 8,428 21,195 30,093 37.882 46,714 71,923 85,121
1922-1951 31,000 MD 6,729 22,630 28,882 33,511 38,151 49,223 54,191
LP3s 7,696 21,412 32,039 41,911 53,664 89,970 110,383
LP3w 8,322 20,320 27,203 32,531 37,968 50,995 56,762
1941-1993 51,000 MD 8,228 24,482 35,428 44,734 55,091 83,797 98,537
LP3s 8,834 23,191 34,655 45,566 58,854 101,657 126,716
LP3w 9,412 22,432 30,629 37,388 44,678 63,853 73,155
1973-1993 51,000 MD 12,757 32,515 46,120 58,233 72,554 119,907 150,941
LP3s 12,985 33,048 45,165 54,813 64,888 89,916 101,378 LP3w 13,184 32,628 43,642 52,079 60,609 80,716 89,471
10,300 27,600 39,600 50,000 62,000 97,800 118,500 Spoon River at Seville. 1636 sq mi
1919-1993 37,300 MD 12,518 24,398 30,069 34,246 38,417 48,264 52,620
LP3s 12,371 24,340 30,906 35,972 41,166 53,830 59,574
LP3w 12,475 24,190 30,352 34,993 39,657 50,672 55,516
Table 16. Continued
Observed Floods at various recurrence intervals, cfs Record Qmax Method 2-year 10-year 25-year 50-year 100-year 500-year 1000-year
1941-1993 36,400 MD 13,429 24,330 29,913 34,119 38,270 48,546 53,104
LP3s 13,370 24,426 30,046 34,217 38,364 48,017 52,211
LP3w 13,504 24,233 29,390 33,099 36,692 44,718 48,068
1973-1993 36,400 MD 14,911 28,500 35,772 41,375 47,158 61,475 68,120
LP3s 14,911 28,822 35,509 40,276 44,849 54,861 58,949
LP3w 14,846 28,935 35,869 40,878 45,738 56,567 61,065
14,000 26,100 32,300 37,800 41,000 51,800 56,000
Sangamon River near Oakford, 5093 sq mi 1931-1993 123,000 MD 23,402 43,319 54,631 65,235 78,745 121,921 144,680
LP3s 22,294 46,445 57,412 64,868 71,719 85,645 90,903
LP3w 21,959 47,156 59,550 68,384 76,824 95,089 102,430
1941-1993 123,000 MD 23,709 44,060 57,930 71,736 88,707 137,921 163,216
LP3s 23,376 46,886 58,229 66,292 73,997 90,736 97,507
LP3w 23,245 47,126 58,995 67,580 75,904 94,407 102,063
1973-1993 68,700 MD 25,848 47,037 60,837 72,286 84,498 115,524 130,104
LP3s 25,889 47,514 60,057 70,114 80,778 108,329 121,561
LP3w 27,133 45,948 54,329 60,124 65,569 71,167 81,794
24,500 46,300 59,000 72,000 87,500 138,000 163,000
Table 16. Concluded
Observed Floods at various recurrence intervals, cfs Record Qmax Method 2-year 10-year 25-year 50-year 100-year 500-year 1000-year
La Moine River at Ripley, 1293 sq mi
1921-1993 28,000 MD 9,411 18,826 23,940 27,913 32,029 42,278 47,049
LP3s 9,385 19,104 23,864 27,274 30,552 37,737 40,669
LP3w 9,359 19,151 24,017 27,532 30,934 38,475 41,586
1941-1993 28,000 MD 10,817 20,414 25,169 28,775 32,471 41,748 46,189
LP3s 10,599 20,831 25,725 29,194 32,504 39,685 42,857
LP3w 10,547 20,924 26,020 29,689 33,236 41,086 44,324
1973-1993 28,000 MD 13,404 24,173 29,697 33,881 38,142 48,516 53,263 LP3s 13,706 24,685 28,046 29,859 31,231 33,289 33,852
LP3w 12,488 27,401 35,422 41,461 47,502 61,581 67,699
12,000 23,000 26,800 31,000 35,000 45,000 49,000
Macoupin Creek near Kane. 868 sq mi
1941-1993 40,000 MD 10,519 22,106 28,196 33,326 39,341 59,392 71,296
LP3s 9,804 23,597 30,441 35,262 39,800 49,324 53,020
LP3w 9,588 24,124 32,113 38,096 44,025 57,530 63,211
1973-1993 27,800 MD 11,482 21,292 26,427 30,348 34,355 44,108 48,541
LP3s 11,283 21,882 26,680 29,972 33,027 39,369 41,821
LP3w 11,062 22,287 27,943 32,071 36,107 4,5201 49,016
10,800 22,800 29,300 34,600 39,600 59,400 71,300
Notes: Suitable T-year flood values are given in bold; MD denotes mixed distribution; LP3s denotes log-Pearson III distribution with sample skew; and LP3w denotes Log-Pearson III distribution with weighted skew
Table 17. Record Periods, Outlier/Inlier Modifications, and Maximum Flood Peaks
Maximum flood Period Outliers/Inliers peak, cfs
Mackinaw River near Green Valley
1922-1993 none 51,000 1922-1951 none 31,000 1941-1993 L1 1,830 → 1,940 51,000
L3 4,400 → 3,757 L4 4,520 → 4,039 L5 4,620 → 4,292
1973-1993 H2 46,700 → 46,106 51,000
Spoon River at Seville
1919-1993 none 37,300 1941-1993 none 36,400 1973-1993 none 36,400
Sangamon River near Oakford
1931-1993 H1 123,000 → 93,289 123,000 H4 45,800 → 49,441 H5 44,700 → 46,807 L5 5,900 → 6,056
1941-1993 H1 123,000 → 95,194 123,000 H4 45,800 → 48,385 H5 44,700 → 45,607 L3 5,940 → 6,239
1973-1993 L2 11,800 → 12,101 68,700
La Moine River at Ripley
1921-1993 L4 3,490 → 3,467 28,000 L5 3,770 → 3,747
1941-1993 none 28,000 1973-1993 none 28,000
Macoupin Creek near Kane
1941-1993 L3 2,640 → 2,268 40,000 1973-1993 H2 26,700 → 26,384 27,800
Notes: H1 denotes the highest observed flood, H2 the second highest flood, and so on. L1 denotes the lowest obeserved flood, L2 the second lowest flood, and so on.
73
Figure 23. Fitted MD and LP3s distributions to observed annual maximum floods: Mackinaw River near Green Valley, 1941-1993
74
Figure 24. Fitted MD and LP3s distributions to observed annual maximum floods: Spoon River at Seville, 1941-1993
75
Figure 25. Fitted MD and LP3s distributions to observed annual maximum floods: Sangamon River near Oakford, 1941-1993
76
Figure 26. Fitted MD and LP3s distributions to observed annual maximum floods: La Moine River at Ripley, 1941-1993
77
Figure 27. Fitted MD and LP3s distributions to observed annual maximum floods: Macoupin Creek near Kane, 1941-1993
78
Mackinaw River near Green Valley, for the 1922-1951 record (Figure 28) clearly demonstrate the
versatility of mixed distribution to fit various shapes of flood frequency curves.
Table 16 provides values of 2-, 10-, 25-, 50-, 100-, 500-, and 1,000-year floods with
mixed distribution (MD), log-Pearson HI with sample skew (LP3s) and log-Pearson HI with
weighted skew (LP3w), for different record lengths at each station. Suitable estimates for design
floods (shown in bold figures in the table) were made for various recurrence intervals (T).
Mackinaw River near Green Valley. The period of record covers 1922 to 1993. Table 16
provides various T-year floods derived using four periods (1922-1993, 1922-1951, 1941-1993,
and 1973-1993). The five highest (H) and lowest (L) floods during these periods are given
below.
Observed H and L floods (cfs)
Period H/L 1 2 3 4 5 1922-1993 H 51,000 46,700 31,000 29,700 26,400
L 1,830 2,420 3,050 3,280 3,370 1922-1951 H 31,00 26,400 24,300 19,700 19,300
L 2,420 3,050 3,280 3,370 3,450 1941-1993 H 51,000 46,700 31,000 29,700 26,400
L 1,830 3,280 4,400 4,520 4,620 1973-1993 H 51,000 46,700 29,700 23,400 23,400
L 1,830 4,730 6,540 6,970 7,830
There is a significant trend of increase in values of both high and low floods over time.
Estimating a representative design flood requires consideration of both trend and suitable
underlying distribution.
Figure 29 plots the values of flood peaks for each of the seven recurrence intervals (2 to
1,000 years) with mixed, LP3s, and LP3w distributions at the middle of each of the four periods
listed above. Curves for 2-, 10-, 25-, and 50-year recurrence intervals show an increasing trend
over time, and the trend becomes more pronounced as T increases. It is considered that the
period (1954 to 1993) adequately represents dry, average, and wet years. It also includes the
recent 20-year period of higher precipitation and increased flows and flood peaks. Thus, suitable
values from these curves for the year 1973 are adopted as design flood peaks (bold type in Table
16).
79
Figure 28. Fitted MD and LP3s distributions to observed annual maximum floods: Mackinaw River near Green Valley, 1922-1951
80
Figure 29. 2- to 1000-year flood peaks for various periods: Mackinaw River
81
Figure 28 shows that for the period 1922-1951, the floods from LP3s for T>10 years will
vary greatly but gradually exceed those from mixed distribution as T increases. Significantly
high positive skews for the periods 1922-1993, 1922-1951, and 1941-1993 cause 100-, 500-, and
1,000-year flood peaks with LP3s to be progressively much higher than with MD. The LP3w
values are lower than the MD values. However, the use of regional skew of -0.4 for the
Mackinaw River basin is open to question, more so when the 72-year record (1922-1993) yields a
sample skew of 0.450. The observed maximum flood of 51,000 cfs may be assigned a recurrence
interval of 50 to 75 years.
Suitable design flood estimates for various recurrence intervals have been derived
considering the representative recent 40-year period (1954-1993), a trend of increasing flood
peak with years, and relative goodness of fit of MD, LP3s, and LP3w curves with the observed
floods.
Spoon River at Seville. The sample skew for the periods 1919-1993, 1941-1993, and
1973-1993 is negative with values of -0.122, -0.206, and -0.444, decreasing with decrease in
length of period of record. Five highest (H) and lowest (L) floods occurring in these periods are
given below. There is no significant trend of increase in values of both high and low floods from
one period to the other.
Observed H and L floods (cfs)
Period H/L 1 2 3 4 5 1919-1993 H 37,300 36,400 34,700 29,200 27,900
L 3,560 3,730 4,530 4,710 5,420 1941-1993 H 36,400 34,700 29,200 27,900 23,700
L 3,730 5,420 5,430 5,630 6,470 1973-1993 H 36,400 34,700 29,200 21,400 21,000
L 3,730 6,470 7,470 7,630 8,840
Figure 30 plots values of flood peaks for each of the seven recurrence intervals with MD,
LP3s, and LP3w for the three periods listed above. These values are plotted at the middle of the
relevant period. The curves show an increase in flood peaks in the 1973-1993 period and the
increase intensifies as T increases. Table 16 provides the best estimates of design floods for 2-,
10-, 25-, 50-, 100-, 500-, and 1,000-year recurrence interval (in bold type) by considering 1954-
1993 as the period representative of dry, average and wet years as well as of recent 20 years of
increasing flows and flood peaks. In developing these estimates, consideration was given to the
82
Figure 30. 2- to 1000-year flood peaks for various periods: Spoon River at Seville
83
representative recent 40-year period, the trend of increasing floods in the last 20 years, and the
relative goodness of fit of MD, LP3s, and LP3w.
Sangamon River near Oakford. The annual flood peak record of 1931-1993 was
analyzed for three periods: 1931-1993, 1941-1993, and 1973-1993, and MD provided the best fit
for the observed floods in these periods. Values of flood peaks for 2-, 10-, and 25-year
recurrence intervals are in the same general range with these distributions, but for T exceeding 25
years and going to 1,000 years, the flood peaks obtained with mixed distributuion become
progressively higher than with LP3s and LP3w for the first two periods, and with LP3w for the
third period (positive sample skew in the third period makes MD values not much higher than
LP3s values).
The five highest (H) and lowest (L) floods occurring in the three periods are given below.
Observed H and L floods (cfs)
Period H/L 1 2 3 4 5 1931-1993 H 123,000 68,700 55,900 45,800 44,700
L 3,480 3,800 4,630 5,670 5,960 1941-1993 H 123,000 68,700 55,900 45,800 44,700
L 3,800 5,670 5,940 8,400 10,000 1973-1993 H 68,700 55,900 45,800 42,900 36,000
L 11,300 11,800 15,800 18,000 18,800
Three of the five lowest floods that occurred in 1931-1993 also occurred in 1941-1993
but none of them occurred during 1973-1993. The highest flood (123,000 cfs) occurred on May
20, 1943. The second highest flood (68,700 cfs) occurred December 12, 1982. The tremendous
difference between the highest and the next highest flood shows that 1943 flood is an outlier and
was detected as such by the general frequency program.
Figure 31 plot values of flood peaks for each of the seven recurrence intervals with MD,
LP3s, and LP3w for the three periods listed above. Table 16 (bold type) provides the best
estimates of design floods for 2-, 10-, 25-, 50-, 100-, 500-, and 1000-year recurrence intervals.
The highest flood (123,000 cfs) has an estimated average recurrence interval of about once in 300
years.
La Moine River at Ripley. The record of three periods was considered: 1921-1993,
1941-1993, and 1973-1993. The five highest (H) and lowest (L) floods in these periods are given
in the following table. 84
Figure 31. 2- to 1000-year flood peaks for various periods: Sangamon River near Oakford
85
Observed H and L floods (cfs)
Period H/L 1 2 3 4 5 1921-1993 H 28,000 27,300 24,100 23,400 22,000
L 1,280 1,800 3,140 3,490 3,700 1941-1993 • H 28,000 27,300 24,100 23,400 22,000
L 1,800 3,490 3,890 3,900 4,290 1973-1993 H 28,000 27,300 23,400 22,000 21,000
L 1,800 4,290 5,430 7,110 8,120
The five highest floods are the same in both 1921-1993 and 1941-1993 periods. Thus,
the period 1921-1940 did not contribute any of these floods. Only the third highest flood
(24,1000 cfs) did not occur in the 1973-1993 period. The lowest floods also increased greatly in
this period. Sample skews for the first two periods (-0.474 and -0.480) are not much different
from the assumed regional skew (-0.4). However, a rapid increase in lowest flood values,
without a corresponding increase in highest values in 1973-1993 period, makes the fitted LP3s
curve flatten out at the upper end, so much so that the 25-year flood peak (28,046 cfs) is slightly
less than 1000-year flood peak (33,852 cfs). However, floods peaks with LP3w for the period
1971-1993 are much higher than LP3s and higher than with MD. Design values (Table 16)
generally follow those developed from mixed distribution. Figure 32 plots values of flood peaks
for each of the seven recurrence intervals with MD, LP3s, and LP3w for the three periods listed
above. There is a trend of increase in flood peaks. Table 16 (bold type) provides the suitable
design flood estimates for the 40-year repesentative period.
Macoupin Creek near Kane. The annual flood record covers the period 1941-1993. The
frequency program was applied to two data sets: 1941-1993 and 1973-1993. The five highest (H)
and lowest (L) floods in these periods are given below.
Observed H and L floods (cfs)
Period H/L 1 2 3 4 5 1941-1993 H 40,000 27,800 26,700 25,400 24,300
L 906 1,410 2,640 2,700 2,730 1973-1993 H 27,800 26,700 19,400 17,500 15,200
L 3,150 3,310 4,910 5,030 7,180
86
Figure 32. 2- to 1000-year flood peaks for various periods: La Moine River at Ripley
87
Only the second and third highest floods in 1941-1993 period occur as the first and
second highest floods in the 1973-1993 period. Moreover, the lowest flood peaks in the 1973-
1993 period are about double or more than those in the 1941-1993 period. Accordingly, the
standard deviation, σ, is greatly reduced from 0.359 for the longer period to 0.261 for 1973-1993
period, which significantly reduces flood peaks with LP3s. These peaks increase slightly with
LP3w, the weighted skew being less negative than the sample skew.
Figure 33 plots values of flood peaks for each of the seven recurrence intervals with MD,
LP3s, and LP3w for the two periods listed above. Table 16 (bold type) provides the best
estimates of design flood for 2-, 10-, 25-, 50-, 100-, 500-, and 1,000-year recurrence intervals.
Peak Flood Versus T Curves
Figure 34 plots design peak floods at recurrence intervals of 2 to 1,000 years for the five
tributaries. Flood peak vs. T curves for the Spoon and La Moine Rivers and Macoupin Creek
show a progressive gradual increase in flood peak as T increases. However, such curves for the
Sangamon and Mackinaw Rivers show a substantial, progressive increase with increase in
recurrence interval. All curves are well defined.
Floods in the Illinois River below Peoria to Grafton are largely governed by outflows
from Peoria Lake and flood flows from the Mackinaw, Spoon, Sangamon, and La Moine Rivers,
and Macoupin Creek, which add considerably to the flood volumes and flood peaks. Concurrent
tributary flood peaks and Illinois River flow below Peoria will primarily determine the flood
hydrographs from Peoria to Grafton.
88
Figure 33. 2- to 1000-year flood peaks for various periods: Macoupin Creek near Kane
89
Figure 34. Flood peak versus T curves for Mackinaw, Spoon, Sangamon, and La Moine Rivers, and Macoupin Creek
90
FLOODS AND STAGES: ILLINOIS RIVER BELOW MARSEILLES
In the river reach from Marseilles to Grafton, there are three long-term U.S.
Geological Survey (USGS) gaging stations where the daily flow, stage, and annual peak flow
data are available: Marseilles, Kingston Mines, and Meredosia.
Record Drainage area River mile Location (years) (sq mi) (mi)
Illinois River at Marseilles 1920-1993 8,259 246.5
Illinois River at Kingston Mines 1941-1993 15,818 144.4
Illinois River at Meredosia* 1939-1993 26,028 71.3
Illinois River at Valley City** 1985-1993 26,742 61.4
Notes:
*stage data for 1939-1993, and flow data for 1939-1989. **stage data for 1985-1993, and flow data for 1990-1993
The general frequency program was used to develop distribution parameter values
(mixed and log-Pearson III distributions) for the entire record as well as split samples to
examine any time trends in various frequency floods and stages at Marseilles, Kingston
Mines, and Meredosia.
Flood Frequency and Design Floods
Annual peak flood data are available at Marseilles, Kingston Mines, and Meredosia for
the 1941-1993 period of record, after adjusting annual peaks at Valley City (1990-1993) to
corresponding values at Meredosia. Table 18 provides 16 top ranked flood peaks at each
station and the date of occurrence.
Flood flows passing through Peoria Lake, on the way to Kingston Mines, are modified
primarily by increased lake storage and also by flood flows in the Mackinaw River entering
the Illinois River at mile 147.7 or 3.3 miles upstream of the gage at Kingston Mines. In order
to investigate the correlation between peak flows at Kingston Mines and at Meredosia, it is
necessary to consider the top five floods at Meredosia and ranks for corresponding floods at
Kingston Mines.
91
Table 18. Ranked Peak Floods at Marseilles, Kingston Mines, and Meredosia (1941-1993)
Illinois River at Marseilles Illinois River at Kingston Mines Illinois River at Meredosia
Rank Date Peak, cfs Date Peak, cfs Date Peak, cfs
1 12/4/82 94,100 12/7/82 88,800 5/26/43 123,000
2 7/14/57 93,900 5/23/43 83,100 3/10/85 122,000
3 5/15/70 92,500 3/6/85 78,800 12/12/82 112,000
4 6/14/81 88,500 3/22/82 77,200 4/19/79 111,000
5 11/29/90 85,800 5/18/70 75,400 6/29/74 110,000
6 2/24/85 84,800 3/21/79 72,300 3/24/82 104,000
7 4/26/50 83,300 5/25/74 71,900 5/2/73 102,000
8 3/14/82 75,800 4/28/50 71,400 4/29/44 102,000
9 2/7/42 74,400 12/3/90 70,800 11/27/85 96,800
10 5/21/43 73,800 3/15/90 69,100 5/22/70 94,800
11 4/6/47 72,000 11/12/85 67,100 3/29/62 91,100
12 3/20/79 71,100 3/9/76 66,300 8/1/93 89,630
13 11/20/85 68,300 4/26/44 64,200 3/37/76 80,600
14 3/6/76 66,900 4/25/73 63,300 2/28/51 77,800
15 5/13/66 66,500 3/25/62 62,100 5/2/50 77,200
16 1/5/93 66,000 2/17/84 60,000 6/7/80 76,900
Over a relatively very long record, corresponding flood peaks at Meredosia and Kingston
Mines will generally have similar recurrence interval (T) values, for T > 20 years.
Flood Frequency Analyses. The general frequency program was used to develop
mixed (MD) and log-Pearson HI (LP3) distribution parameters for the entire record, as well as
split samples to examine any time trends in floods at the three stations. Table 19 provides the
results, which show the change in parameter values when different record lengths are used to
fit frequency distributions to observed data.
Table 20 presents the flood peaks derived with the fitted parameters for recurrence
intervals of 2, 10, 25, 50, 100, 500, and 1,000 years for various record lengths at each of the
three stations for mixed distribution (MD), log-Pearson III with sample skew (LP3s)
distribution, and log-Pearson III with weighted skew (LP3w) distribution. The outlier/inlier
detection and modification resulted only in minor changes to one flood each in 1904-1993,
1941-1993, and 1941-1980 records for the Illinois River at Marseilles, and Table 21 gives
relevant information.
Design floods. Figures 35-37 show flood frequency curves fitted using mixed
distribution or MD (together with two component normal distributions) and log-Pearson III
with sample skew (LP3s) distribution to the observed annual floods for the period 1969-1993
at Marseilles, Kingston Mines, and Meredosia. The mixed distribution yields an S-shaped
curve fitting the observed flood peaks very well compared to a rather poor fit with LP3s
distribution, at least for Marseilles and Kingston Mines. However, if 1941-1993 data are
used, the fitted MD curves have a reduced severity of reversed curvature but still fit the data
somewhat better than with LP3s distribution (Figures 38-40).
Design floods for T = 2, 10, 25, 50, 100, 500, and 1,000 years at Marseilles, Kingston
Mines, and Meredosia were determined using the same procedure used for major tributaries
(see previous chapter). This procedure allows consideration of any time trends and estimation
of design values as indicated for the 1954-1993 period. Design T-year flood peaks are given
in bold in Table 20.
93
Rank, Meredosia flood
Rank, corresponding flood at Kingston Mines
1
2
2
3
3
1
4
6
5
7
Table 19. Fitted Mixed Distribution (MD) and Log-Pearson III (LP3s) Distribution Parameters for Peak Floods
MD parameters LP3s parameters
Period a µ1 σ1 µ2σ2µσ g
Illinois River at Marseilles
1904-1993 0.759 4.584 0.166 4.826 0.090 4.643 0.183 -0.211
1904-1940 0.640 4.479 0.139 4.694 0.115 4.556 0.166 -0.059
1941-1993 0.526 4.613 0.167 4.803 0.110 4.703 0.171 -0.446
1941-1980 0.071 4.340 0.100 4.710 0.134 4.684 0.163 -0.532
1969-1993 0.357 4.551 0.100 4.843 0.093 4.739 0.170 -0.394
1981-1993 0.318 4.534 0.123 4.867 0.095 4.761 0.187 -0.641
Illinois River at Kingston Mines
1941-1993 0.702 4.611 0.147 4.817 0.064 4.672 0.159 -0.381
1941-1980 0.850 4.622 0.153 4.787 0.025 4.647 0.153 -0.310
1969-1993 0.367 4.550 0.104 4.824 0.070 4.724 0.157 -0.623
1981-1993 0.376 4.627 0.181 4.826 0.075 4.752 0.158 -1.070
Illinois River at Meredosia
1921-1993 0.315 4.615 0.187 4.856 0.122 4.780 0.184 -0.692
1921-1940 0.527 4.602 0.236 4.809 0.110 4.700 0.214 -0.677
1941-1993* 0.140 4.542 0.086 4.855 0.125 4.811 0.162 -0.413
1941-1980 0.299 4.671 0.157 4.851 0.121 4.797 0.156 -0.428
1969-1993 0.216 4.634 0.100 4.923 0.102 4.861 0.157 -0.599
1981-1993 0.485 4.738 0.185 4.962 0.089 4.853 0.182 -0.742
Notes: MD = mixed distribution (µ1 and σ1 are mean and standard deviation of one component distribution; µ2 and σ2 of second distribution; and a is weight of first component distribution and 1-a of second component distribution) LP3s = log-Pearson Type III distribution with sample skew (µ, σ, and g denote mean, standard deviation, and skew) * = fifth window parameters (see general frequency program)
Table 20. Flood Frequency Analysis for Illinois River at Marseilles, Kingston Mines, and Meredosia
Observed Floods at various recurrence intervals, cfs Record Qntax Method 2-year 10-year 25-year 50-year 100-year 500-year 1000-year
Illinois River at Marseilles 1904-1993 94,100 MD 44,544 75,177 86,315 93,919 101,253 118,375 126,249
LP3s 44,581 74,684 89,147 99,616 109,837 133,014 142,854 LP3w 44,660 74,581 88,799 99,031 108,971 131,357 140,762
1904-1940 77,000 MD 36,102 59,153 69,010 75,860 82,407 96,967 103,082 LP3s 36,131 58,647 69,811 78,061 86,262 105,420 113,799 LP3w 36,796 57,841 67,090 73,489 79,516 92,455 97,669
1941-1993 94,100 MD 52,704 81,009 93,029 101,539 109,833 128,872 137,227 LP3s 51,950 81,746 94,350 102,876 110,759 127,182 133,606 LP3w 51,851 81,886 94,765 103,543 111,710 128,888 135,673
1941-1980 93,900 MD 49,783 75,244 87,198 95,879 104,392 123,992 132,446 LP3s 49,910 76,041 86,504 93,383 99,601 112,112 116,843 LP3w 49,584 76,503 87,838 95,504 102,592 117,381 123,177
1969-1993 94,100 MD 59,447 86,520 96,784 103,840 110,504 125,128 131,189 LP3s 56,187 88,625 102,648 112,252 121,224 140,221 147,773 LP3w 56,208 88,597 102,561 112,111 121,020 139,848 147,318
1981-1993 94,100 MD 64,512 92,830 103,945 111,630 118,922 134,966 141,619 LP3s 60,430 96,566 110,788 119,985 128,169 144,192 150,070 LP3w 59,245 98,376 116,068 128,440 140,179 165,612 175,939
55,000 84,200 96,000 105,000 113,000 131,000 137,000
Table 20. Continued
Observed Floods at various recurrence intervals, cfs Record Qmax Method 2-year 10-year 25-year 50-year 100-year 500-year 1000-year
Illinois River at Kingston Mines
1941-1993 88,800 MD 48,862 73,025 80,617 85,925 91,268 105,454 112,726
LP3s 48,126 73,910 84,927 92,445 99,450 114,245 120,117
LP3w 48,162 73,864 84,787 92,217 99,124 113,654 119,398
1941-1980 83,100 MD 45,312 65,851 75,470 84,205 92,892 113,263 122,245
LP3s 45,123 68,600 78,834 85,903 92,561 106,868 112,645
LP3w 45,312 68,358 78,091 84,700 90,837 103,736 108,834
1969-1993 88,800 MD 58,849 78,498 85,487 90,168 94,539 103,881 107,667
LP3s 54,922 81,483 91,548 97,977 103,653 114,670 118,687
LP3w 54,200 82,510 94,452 102,546 110,050 125,773 131,965
1981-1993 88,800 MD 61,554 81,772 90,219 96,550 103,306 123,929 135,570
LP3s 60,167 84,705 91,885 95,839 98,922 103,839 105,293
LP3w 57,460 88,906 102,857 112,575 121,785 141,765 149,904
52,200 76,500 85,000 91,800 98,200 113,000 117,500
Illinois River at Meredosia
1921-1993 123,000 MD 63,998 97,815 112,800 123,509 133,974 157,894 168,260
LP3s 63,284 99,501 113,289 122,048 129,730 144,433 149,706
LP3w 62,826 100,241 115,302 125,178 134,069 151,804 158,438
Table 20. Concluded
Observed Floods at various recurrence intervals, cfs Record Qmax Method 2-year 10-year 25-year 50-year 100-year 500-year 1000-year
1921-1940 105,000 MD 54,616 86,240 101,372 113,829 128,088 170,959 192,840
LP3s 52,925 89,849 104,676 114,296 122,862 135,594 145,704
LP3w 51,674 91,926 110,776 124,132 136,918 164,926 176,397
1941-1993 123,000 MD 67,546 100,900 116,003 126,834 137,384 161,378 171,668
LP3s 66,442 102,411 117,628 127,946 137,511 157,539 165,419
LP3w 66,409 102,457 117,764 128,163 137,820 158,091 166,086
1941-1980 123,000 MD 64,542 96,591 110,999 121,276 131,294 153,960 163,642
LP3s 64,264 97,374 111,150 120,419 128,962 146,697 153,619
LP3w 64,178 97,492 111,495 120,970 129,742 148,083 155,292
1969-1993 122,000 MD 77,206 109,420 122,983 132,468 141,529 161,675 170,109
LP3s 75,201 111,883 125,952 134,998 143,030 158,753 164,536
LP3w 74,319 113,126 129,495 140,590 150,875 172,425 180,912
1981-1993 122,000 MD 78,182 110,037 127,051 137,159 147,371 174,128 188,093
LP3s 75,090 116,781 132,185 141,812 150,141 165,753 171,233
LP3w 73,069 119,893 141,037 155,829 169,877 200,367 212,755
71,000 105,000 120,600 130,000 139,500 161,000 169,000
Note: Numbers in bold type represent design T-year flood peaks.
Table 21. Record Periods, Outlier/Inlier Modifications, and Maximum Flood Peaks
Period Outlier/Inliers Maximum Flood Peak, cfs
Illinois River at Marseilles 1904-1993 H1 94,100 → 98,097 94,100 1904-1940 none 77,000 1941-1993 H1 94,100 → 96,018 94,100 1941-1980 L2 19,300 → 19,568 93,900 1969-1993 none 94,100 1981-1993 none 94,100
Illinois River at Kingston Mines 1941-1993 none 88,800 1941-1980 none 83,100 1969-1993 none 88,800 1981-1993 none 88,800
Illinois River at Meredosia 1921-1993 none 123,000 1921-1940 none 105,000 1941-1993 none 123,000 1941-1980 none 123,000 1969-1993 none 122,000 1981-1993 none 122,000
Note: H1 is the highest flood and L2 is the second lowest flood during the record considered.
98
Figure 35. Fitted MD and LP3s distribution to annual maximum floods, Illinois River at Marseilles, 1969-1993
99
Figure 36. Fitted MD and LP3s distribution to annual maximum floods, Dlinois River at Kingston Mines, 1969-1993
100
Figure 37. Fitted MD and LP3s distribution to annual maximum floods, Illinois River at Meredosia, 1969-1993
101
Figure 38. Fitted MD and LP3s distribution to annual maximum floods, Illinois River at Marseilles, 1941-1993
102
Figure 39 . Fitted MD and LP3s distribution to annual maximum floods, Illinois River at Kingston Mines, 1941-1993
103
Figure 40. Fitted MD and LP3s distribution to annual maximum floods, Illinois River at Meredosia, 1941-1993
104
Figure 41 plots design peak floods at recurrence intervals of 2 to 1,000 years at
Marseilles, Kingston Mines, and Meredosia. Similar T-year peak floods at Kingston Mines
are significantly lower than at Marseilles, largely due to Peoria Lake flood surcharge storage
capacity that lowers the flood peak but increases the number of days with high flows.
Stage Frequency and Design Stages
Annual peak stage data are available at Marseilles, Kingston Mines, and Meredosia for
the 1941-1993 period. To investigate any time trends, the data were analyzed for the entire
period and subperiods: 1941-1993,1941-1980, 1969-1993, and 1981-1993. The datum of the
USGS gage at Marseilles is 462.91 ft-msl (feet above mean sea level), 428.00 ft-msl at
Kingston Mines and 418.00 ft-msl at Meredosia.
The general frequency program was used to develop mixed and log-Pearson HI
distribution parameters for the entire record as well as for subperiods. Table 22 shows the
change in parameter values when different record lengths are used to fit stage frequency
distributions to the observed data.
Table 23 gives the stage peaks derived with the fitted parameters for recurrence
intervals of 2, 10, 25, 50, 100, 500, and 1,000 years for various record lengths at each station
for mixed distribution and log-Pearson III distribution with sample skew (LP3s) and weighted
skew (LP3w). The outlier/inlier detection and modification resulted in the following minor
adjustments.
Station Record Outliers/inliers Maximum stage, feet
Meredosia 1941-1993 H5 26.75→26.70 28.61 Meredosia 1941-1980 H3 26.75→26.66 28.61 Meredosia 1969-1993 H1 27.62→28.12 27.62 Kingston Mines 1941-1993 H5 24.28→24.03 26.02
Note: H1 denotes the highest stage, H2 the second highest stage, and so on for a particular record; no outliers/inliers detected in the Marseilles stage records
Design T-year stages are given in bold in Table 23. Corresponding water surface
elevations above mean sea level (feet-msl) are also given. Figures 42 and 43 plot design peak
stages and water elevations at recurrence intervals of 2 to 1,000 years at Marseilles, Kingston
105
Figure 41. T-year flood peaks for Illinois River at Marseilles, Kingston Mines, and Meredosia
106
Table 22. Fitted Mixed Distribution (MD) and Log-Pearson HI Distribution (LP3s) Parameters for Peak Stages
MD parameters LP3s parameters Period a µ1 σ1 µ2 σ2 µ a g
Dlinois River at Meredosia
1941-1993 0.131 1.041 0.072 1.292 0.089 1.259 0.122 -0.606 1941-1980 0.255 1.112 0.105 1.292 0.079 1.246 0.116 -0.657 1969-1993 0.250 1.175 0.111 1.337 0.084 1.296 0.115 -0.593 1981-1993 0.181 1.069 0.083 1.350 0.078 1.299 0.134 -0.916
Dlinois River at Kingston Mines
1941-1993 0.859 1.258 0.080 1.189 0.122 1.248 0.090 -0.353 1941-1980 0.913 1.240 0.079 1.180 0.145 1.235 0.088 -0.341 1969-1993 0.908 1.294 0.072 1.107 0.005 1.277 0.088 -0.299 1981-1993 0.829 1.318 0.061 1.147 0.058 1.288 0.088 -0.613
Illinois River at Marseilles
1941-1993 0.449 0.918 0.128 1.084 0.078 1.009 0.132 -0.604 1941-1980 0.288 0.869 0.128 1.045 0.082 0.994 0.126 -0.753 1969-1993 0.313 0.874 0.078 1.109 0.072 1.036 0.132 -0.520 1981-1993 0.596 0.981 0.141 1.167 0.038 1.056 0.146 -0.741
Notes: MD = mixed distribution (µ1 and σ1 are mean and standard deviation of one component distribution; µ2 and
σ2 of second component distribution; and a is weight of first component distribution and 1-a of second component distribution)
LP3s = Log-Pearson Type IE distribution with sample skew (µ, σ, and g denote mean, standard deviation, and skew)
107
Table 23. Stage Frequency Analysis for Illinois River at Marseilles, Kingston Mines, and Meredosia
Observed Stages at various recurrence intervals, feet Record max. stage Method 2-year 10-year 25-year 50-year 100-year 500-year 1000-year
Illinois River at Marseilles
1941-1993 16.78 MD 10.73 14.52 15.99 17.00 17.95 20.07 20.97
LP3s 10.53 14.72 16.27 17.25 18.11 19.76 20.36
LP3w 10.46 14.82 16.52 17.63 18.64 20.66 21.43
1941-1980 15.20 MD 10.31 13.68 15.05 16.00 16.89 18.86 19.67
LP3s 10.23 13.86 15.09 15.83 16.46 17.61 18.00
LP3w 10.09 14.05 15.58 16.59 17.49 19.30 19.99
1969-1993 16.78 MD 11.64 15.31 16.68 17.60 18.47 20.32 21.08
LP3s 11.14 15.68 17.41 18.53 19.54 21.53 22.27
LP3w 11.07 15.76 17.66 18.92 20.08 22.46 23.39
1981-1993 16.78 MD 12.73 16.04 17.12 18.08 19.43 23.51 25.31
LP3s 11.85 16.90 18.67 19.73 20.68 22.39 22.98
LP3w 11.60 17.26 19.66 21.30 22.82 26.06 27.34
Design stages, ft 11.13 14.96 16.40 17.52 18.25 20.14 20.95
Design water elevation, ft-msl 474.04 477.87 479.31 480.43 481.16 483.05 483.86
Illinois River at Kingston Mines
1941-1993 26.02 MD 17.83 22.84 25.00 26.52 28.00 31.38 32.88
LP3s 17.92 22.90 24.81 26.06 27.19 29.49 30.38
LP3w 17.94 22.88 24.75 25.97 27.07 29.28 30.12
Table 23. Continued
Observed Stages at various recurrence intervals, feet Record max. stage Method 2-year 10-year 25-year 50-year 100-year 500-year 1000-year
1941-1980 26.02 MD 17.27 22.01 24.11 25.62 27.15 31.09 33.18
LP3s 17.38 22.10 23.92 25.11 26.19 28.38 29.22
LP3w 17.41 22.07 23.83 24.97 26.00 28.06 28.84
1969-1993 25.55 MD 19.26 24.13 26.14 27.52 28.82 31.63 32.79
LP3s 19.10 24.31 26.35 27.69 28.91 31.43 32.41
LP3w 19.17 24.24 26.15 27.38 28.48 30.69 31.52
1981-1993 25.55 MD 20.04 24.53 26.29 27.48 28.59 30.96 31.92
LP3s 19.84 24.78 26.47 27.50 28.40 30.08 30.68
LP3w 19.67 24.97 26.98 28.30 29.48 31.88 32.79
Design stages, ft 18.50 23.40 25.50 26.90 28.05 31.10 32.50
Design water elevation, ft 446.50 451.40 453.50 454.90 456.05 459.10 460.50 msl
Illinois River at Meredosia
1941-1993 28.61 MD 18.84 25.08 27.71 29.55 31.28 35.11 36.70
LP3s 18.68 25.42 27.85 29.38 30.72 33.29 34.21
LP3w 18.57 25.56 28.25 29.99 31.56 34.70 35.88
1941-1980 28.61 MD 18.31 24.00 26.29 27.86 29.34 32.51 33.85
LP3s 18.15 24.25 26.39 27.71 28.85 30.98 31.74
LP3w 17.98 24.47 26.96 28.59 30.05 33.00 34.11
Table 23. Concluded
Observed Stages at various recurrence intervals, feet Record max. stage Method 2-year 10-year 25-year 50-year 100-year 500-year 1000-year
1969-1993 27.62 MD 20.37 26.96 29.68 31.55 33.32 37.17 38.77
LP3s 20.31 27.17 29.64 31.19 32.55 35.15 36.09
LP3w 20.14 27.39 30.23 32.11 33.81 37.28 38 61
1981-1993 27.62 MD 21.27 27.57 30.12 31.86 33.50 37.07 38.54
LP3s 20.85 28.31 30.64 31.98 33.07 34.93 35.52
LP3w 20.23 29.19 32.95 35.51 37.90 42.94 44.95
Design stages, ft 19.50 25.82 28.38 30.08 31.55 34.76 36.05 Design water elevation, ft msl 437.50 443.82 446.38 448.08 449.55 452.76 454.05
Note: Numbers in bold type represent design T-year stages.
Figure 42. T-year stages for Illinois River at Marseilles, Kingston Mines, and Meredosia
111
Figure 43. T-year water elevations for Illinois River at Marseilles, Kingston Mines, and Meredosia
112
Mines, and Meredosia. Design elevations at Meredosia are 0.78 to 1.00 feet higher than those
derived in the chapter on "Water Stages and Elevations in Alton Pool, Illinois River," for the
1941-1993 record because of the consideration of time trends in increasing stages.
Concurrency of Tributary and Illinois River Flood Peaks
Table 24 provides the USGS streamgaging stations starting from the Illinois River at
Marseilles downstream to Grafton and stations (nearest to Illinois River) on major tributaries
entering the river in this reach. It includes information on USGS station number, length of
historical record of daily flows, drainage area in square miles, tributary station distance
upstream of confluence with the Illinois River, drainage area of tributaries at their mouth, and
Dlinois River miles for stations along the river (Marseilles, Kingston Mines, and Meredosia)
and at the confluence of tributaries with the river. Only annual peak flow data are available
for the period 1957-1988 for the Mackinaw River near Green Valley. Consequently, daily
flow hydrographs for this period cannot be developed.
Table 25 provides annual flood peaks and dates of occurrence for 1941-1993 record
ten highest peak flood events (ranked 1 to 10 in descending order of magnitude) for the
Illinois River at Meredosia. Flood ranks, magnitudes, and dates of occurrence at other
stations in corresponding years are also included. To consider concurrency of tributary and
Illinois River flood peaks, the lag time in days between stations, and the effect of Peoria Lake,
flood data and available daily flow data were analyzed and evaluated in terms of two reaches:
Marseilles to Kingston Mines and Kingston Mines to Meredosia.
In the Marseilles to Kingston Mine reach, the Fox River at Dayton is 5.3 miles
upstream of the Illinois River, and the confluence is only 6.9 miles below the Marseilles gage.
For all practical purposes, the lag between these flows is negligible. The Vermilion River
near Lenore is 17.2 miles upstream of confluence and 20.2 miles downstream of the
Marseilles gage. Travel times to the confluence will not be significantly different. Peoria
Lake stretches from about mile 182 to 162 (starting 64.5 miles downstream of the Marseilles
gage), about 44 miles below the confluence of the Vermilion and Illinois Rivers. This may
introduce a lag of one or two days during flood conditions. Travel through Peoria Lake
includes both translation and attenuation of the flood peak. However, when the crest of the
113
Table 24. Relevant Data for the Illinois River and Major Tributaries below Marseilles
Drainage Drainage USGS Historical area Upstream of area at mouth River
Tributary station no. record (sq mi) mouth (mi) (sq mi) mile
Illinois River 05543500 1920-1993 8259 246.5 at Marseilles
Fox River 05552500 1916-1993 2642 5.3 2658 239.6 at Dayton
Vermilion River near Leonore 05555300 1972-1993 1251 17.2 1331 226.3 at Lowell 05555000 1932-1971 1278 10.5 1331 226.3
Mackinaw River 05568000 1922-1956 1073 17.3 1136 147.7 near Green Valley 1957-
1988* 1989-1993
Illinois River 05568500 1940-1993 15818 144.4 at Kingston Mines
Spoon River 05570000 1919-1993 1636 38.7 1855 120.4 at Seville
Sangamon River 05583000 1931-1993 5093 25.7 5419 88.9 near Oakford
LaMoine River 05585000 1921-1993 1293 12.3 1350 83.5 at Ripley
Illinois River at Meredosia 05585500 1939-1989 26028 71.3 at Valley City 05586100 1990-1993 26742 61.4
Macoupin Creek 05587000 1941-1993 868 16.1 961 23.2 near Kane
Note: An asterisk denotes data for annual maximum peak flow only.
114
Table 25. Peak Floods in the Illinois River and Major Tributaries below Marseilles
///. River Ill. River Kingston Peoria Ill . River
Macoupin Meredosia La Moine Sangamon Spoon Mines Mackinaw Lake Vermilion Fox Marseille s
Rank 1 1 18 1 28 2 11 10 38 10 Qp 40,000 123,000 14,500 123,000 12,900 83,100 18,200 21,400 10,600 73,800 Date 5/18/43 5/26/43 5/21/43 5/20/43 5/20/43 5/23/43 5/19/43 5/21/43 5/21/43 5/21/43
Rank 10 2 1 21 3 3 10 18 4 6 Qp 19,400 122,000 28,800 26,500 29,200 78,800 18,400 18,500 28,400 84,800 Date 2/24/85 3/10/85 3/7/85 2/25/85 3/6/85 3/6/85 2/25/85 2/25/85 3/5/85 2/24/85
Rank 3 3 6 2 9 1 1 2 6 1 Qp 26,700 112,000 21,000 68,700 21,000 88,800 51,000 31,800 26,000 94,100 Date 12/4/82 12/12/82 12/5/82 12/5/82 4/4/83 12/7/82 12/6/82 12/4/82 12/3/82 12/4/82
Rank 2 4 34 3 30 6 6 12 2 12 Qp 27,800 111,000 8,120 55,900 12,600 72,300 23,600 21,200 29,800 71,100 Date 4/12/79 4/19/79 4/12/79 4/15/79 3/31/79 3/24/79 3/5/79 3/5/79 3/20/79 3/20/79
Rank 22 5 29 6 1 7 30 19 5 19 Qp 12,800 110,000 10,000 42,900 36,400 71,900 7,910 17,400 26,800 64,000 Date 1/21/74 6/29/74 6/2/74 6/25/74 6/24/74 5/25/74 6/5/74 6/23/74 5/17/74 5/22/74
Rank 31 6 28 29 17 4 8 8 25 8 Qp 9,140 104,000 10,700 23,700 16,900 77,200 20,000 23,000 14,800 75,800 Date 2/21/82 3/24/82 3/16/82 3/20/82 7/21/82 3/22/82 3/12/82 4/13/82 3/13/82 3/14/82
Rank 20 7 22 4 21 14 4 21 11 17 Qp 13,300 102,000 12,400 45,800 16,000 63,300 29,700 16,500 18,300 64,900 Date 4/23/73 5/2/73 4/22/73 4/25/73 4/24/73 4/25/73 4/21/73 12/31/72 4/23/73 1/1/73
Table 25. Concluded
Ill. River Ill. River Kingston Peoria Ill. River
Macoupin Meredosia La Moine Sangamon Spoon Mines Mackinaw Lake Vermilion Fox Marseilles
Rank 4 8 9 5 18 13 5 9 13 23 Qp 25,400 102,000 17,100 44,700 16,600 64,200 26,400 22,300 18,000 56,100 Date 4/24/44 4/29/44 4/25/44 4/26/44 3/17/44 4/26/44 4/24/44 4/24/44 3/5/44 4/24/44
Rank 24 9 5 19 23 11 9 15 12 13 QP 12,100 96,800 22,000 28,400 15,000 67,100 19,200 20,200 18,100 68,300 Date 12/12/85 11/27/85 11/19/85 11/25/85 11/20/85 11/12/85 11/21/85 11/19/85 11/10/85 11/20/85
Rank 8 10 3 7 5 5 33 4 15 3 QP 20,100 94,800 24,100 42,300 23,700 75,400 7,000 24,800 17,000 92,500 Date 10/14/69 5/22/70 9/27/70 5/3/70 5/17/70 5/18/70 9/22/70 5/15/70 6/3/70 5/15/70
Note: Qp denotes the peak flood in cubic feet per second (cfs)
inflow flood hydrograph covers many days, attenuation of the flood peak is mostly attributable
to lake surcharge storage and the translation time causing lag between inflow and outflow
flood peaks may be a day or two depending on the flood magnitude. The Mackinaw River
near Green Valley is 17.3 miles upstream of the Illinois River confluence, which is about 14
miles below Peoria Lake and 3.3 miles upstream of the Kingston Mines gage. It may take a
day or less for floods in Mackinaw River near Green Valley to travel to the confluence.
Travel time from Marseilles to Kingston Mines, a distance of 81 miles excluding Peoria Lake,
may be two to three days during high flood conditions plus a lag of one to two days caused by
Peoria Lake. Thus, the total travel time or lag from Marseilles to Kingston Mines may be
three to five days during major floods.
In the Kingston Mines to Meredosia reach, the first confluence is with the Spoon River
at river mile 120.4, or 24 miles downstream of Kingston Mines. During flood flows, it may
take 12 to 24 hours to travel this distance. The Spoon River at Seville is 38.7 miles upstream
of the Illinois River, and it may take a day or more for the flood to travel from Seville to the
Illinois River. The next confluence is with the Sangamon River at river mile 88.9, a distance
of 31.5 miles that may entail lags of about one day during flood conditions. The Sangamon
River at Oakford is 25.7 miles upstream of its mouth, and the flood may take about a day to
reach the confluence. The LaMoine River meets the Illinois River atriver mile 83.5, only 5.4
miles below the confluence of the Illinois and Sangamon Rivers. The LaMoine River at
Ripley is only 12.3 miles upstream of its mouth at the Illinois River. The Meredosia station is
at river mile 71.3. Lag time from Kingston Mines to Meredosia may range from two to four
days during high flood episodes. When the two reaches are combined, flood flows from
Marseilles may take five to nine days to reach Meredosia.
Concurrency and the other effects are discussed below for the top five floods at
Meredosia. Table 26 provides daily flow peak or values at the gaging stations along the
Illinois River and major tributaries entering the river between Marseilles and Meredosia. The
denotes the maximum daily peak flow. The lag or the number of days occurs at a
station upstream before occurs at Meredosia are also included together with the dates on
which occurred. Values of t1(t2) and t3(t4) for the daily flow hydrograph at each station are
also shown.
117
Table 26. Relevant Information for Top Five Floods: Marseilles to Meredosia
Meredosia Kingston Peoria rank Item Meredosia LaMoine Sangamon Spoon Mines Lake Vermilion Fox Marseilles
1 123,000 14,200 120,000 10,600 82,200 19,900 9,960 70,700 Date 5/27/43 5/21/43 5/20/43 5/20/43 5/23/43 5/21/43 5/21/43 5/21/43 Lag 6 7 7 4 6 6 6 t1(t2) 11(6) 2(1) 1(1) 2(2) 3(2) 1(1) 1(1) 2(1) t3(t4) 15(7) 5(4) 2(1) 2(2) 10(5) 1(1) 1(1) 3(2)
2 120,000 26,800 25,900 27,800 77,800 13,000* 24,000 70,300* Date 3/10/85 3/7/85 2/25/85 3/6/85 3/6/85 3/5/85 3/5/85 3/5/85 Lag 3 13 4 4 5 5 5
t1(t2) 10(3,4,6) 1(1) 6(1) 1(1) 4(2) 1(1) 2(2) 1(1) t3(t4) 16(6,7,9) 2(2) 11(2) 2(1) 8(3) 1(1) 2(2) 2(1)
17,000 77,000
Date 2/25/85 2/24/85 Lag 13 14 t1 (t2) 1(1) 2(1) t3 (t4) 1(1) 2(1)
3 112,000 19,000 63,200 19,100 86,700 30,000 22,800 87,800
Date 12/12/82 12/6/82 12/5/82 12/5/82 12/7/82 12/4/82 12/3/82 12/4/82 Lag 6 7 7 5 8 9 8
t1(t2) 5(4)+2 1(1) 2(1) 1(1) 4(2) 1(1) 1(1) 2(1) t3(t4) 12(4) 2(1) 2(1) 2(2) 6(2) 2(1) 2(1) 3(1)
4 109,000 7,460 54,200 10,300* 65,500* 16,400 17,000* 62,700* Date 4/19/79 4/12/79 4/15/79 4/8/79 4/15/79 4/13/79 4/12/79 4/13/79 Lag 7 4 6 4 6 7 6 t1(t2) 6(2) 5(1) 2(1) 2(1) 3(2) 2(2) 1(1) 1(1)
Table 26. Concluded
Meredosia Kingston Peoria rank Item Meredosia LaMoine Sangamon Spoon Mines Lake Vermilion Fox Marseilles
t3(t4) 10(3) 5(1) 4(2) 3(2) 7(3) 2(2) 2(1) 1(1)
12,100 72,200 28,900 68,500
Date 3/31/79 3/24/79 3/20/79 3/20/79 Lag 19 26 27 30 t1(t2) 2(1) 15(5) 2(1) 1(1) t3(t4) 3(1) 21(7) 2(1) 2(1)
5 110,000 6,230* 42,800 32,700 69,800* 16,200 10,400* 34,600* Date 6/29/74 6/26/74 6/25/74 6/24/74 6/24/74 6/23/74 6/23/74 6/23/74 Lag 3 4 5 5 6 6 6 t1(t2) 6(3) 2(2) 1(1) 2(1) 3(2) 1(1) 1(1) 2(1) t3(t4) 10(5) 3(2) 3(2) 2(1) 4(2) 1(1) 2(2) 2(1)
9,860 71,900 21,600 57,800 Date 6/2/74 5/25/74 5/17/74 5/23/74 Lag 27 35 43 37 t1(t2) 2(2) 7(5) 2(1) 2(2) t3(t4) 4(3) 10(6) 2(1) 2(2)
Notes: * = relevant daily flow peak from a nearby hydrograph = daily flow peak, in cfs, in a water year October to September
t1 = number of days daily flow > 0.9 t2 = number of day with in t1 period
t3 = number of days with daily flow > 0.8 t4 = number of day with in t3 period Lag = date of at Meredosia - date of at station under consideration
t1 = number of days daily flow > 0.9
t2 = number of day in the t1 period when occurs. There can be more than one value
of t2 if more than 1 day have the same
t3 = number of days daily flow > 0.8
t4 = number of day in the t3 period when occurs. There can be more than one value
of t4 if more than one day has the same
When at a tributary station, Kingston Mines or Marseilles, occurred much earlier
or much later than at Meredosia, another relevant value from a somewhat lower flood rise was
used. This occurred during multiple peak hydrographs.
1. Flood of May 1943. Daily flow peak or (123,000 cfs) at Meredosia occurred on
May 27, 1943. Such flows occurred at Kingston Mines (82,200 cfs) and at the Sangamon
River at Oakford (120,000 cfs) on May 23 and May 20, respectively, or four days and seven
days prior to the peak flow at Meredosia. The confluence of the Sangamon and Illinois Rivers
is 55.5 miles downstream of Kingston Mines and only 17.6 miles upstream of Meredosia.
Allowing for a day or less of travel time from Oakford to the confluence, if the peak flow had
occurred May 23 or 24 at Oakford (so that flood peaks in the Illinois River and the Sangamon
River are practically concurrent), the peak at Meredosia would have probably been 150,000
cfs or more, but the crest segment of the flood hydrograph would have been about two-thirds
the width. Nonconcurrent peak flow contributions, particularly from the Kingston gage and
the Sangamon River (and to a lesser extent from the Spoon and LaMoine Rivers) results in an
attenuated and wider crest of the flood hydrograph at Meredosia. Hydrographs at various
stations are shown in Figure 44, for which day 1 starts on May 11, 1943.
From Marseilles to Kingston Mines, daily flow peaks occurred at Marseilles, in the
Fox and Vermilion Rivers, and at Kingston Mines on May 21, May 21, May 21, and May 23,
respectively. The double-peak hydrograph at Marseilles, combined with those from the Fox
and Vermilion Rivers, was modified by Peoria Lake. A daily flow peak of 16,300 cfs
occurred on May 19 in the Mackinaw River near Green Valley, and it caused a minor peak
120
Figure 44. Flood hydrographs at Illinois River and tributary gaging stations, May 1943 flood
121
(73,000 cfs) in the daily flow hydrograph at Kingston Mines on May 20 followed by a of
82,200 cfs on May 23.
2. Flood of March 1985. The of 120,000 cfs at Meredosia occurred on March 10,
1985, and that at Kingston Mines (77,800 cfs) occurred on March 6, or four days earlier than
at Meredosia. This time difference or lag is the same as for the May 1943 flood. The Spoon,
Sangamon, and LaMoine River daily flow peaks have ranks of 3, 21, and 1, respectively, in
the 1941-1993 record. The daily flow at Oakford exceeded 21,060 cfs from February 24 to
March 6, with one peak of 25,900 cfs on February 25 and another of 23,400 cfs on March 5.
Thus the flood peaks at Kingston Mines and in the Spoon, Sangamon, and LaMoine Rivers
are rather in phase timewise, and produced a second high at Meredosia. A very flat crest
segment of the Sangamon River flow hydrograph significantly increased the width of the crest
segment at Meredosia.
From Marseilles to Kingston Mines, values of concurrent (or close in value to
actual ) occurred March 5 at Marseilles and March 6 at Kingston Mines. The hydrograph
at Kingston Mines (Figure 45) was significantly affected by the distinctly double-peak
hydrograph at Marseilles, though the time difference between the two peaks of Marseilles
hydrograph was significantly reduced by the flood flow passing through Peoria Lake on the
way to Kingston Mines.
3. Flood of December 1982. The of 112,000 cfs at Meredosia occurred on
December 12, 1982, and that at Kingston Mines (86,700 cfs) occurred on December 7, i.e.,
five days earlier than at Meredosia. Flood peaks on the Spoon, Sangamon, and LaMoine
Rivers have ranks of 9, 2, and 6, respectively, in the 1941-1993 record, and corresponding
values occurred on December 5, 5, and 6, all a day or two earlier than at Kingston Mines.
Allowing for travel time for these tributary flows and flow at Kingston Mines, the daily flow
hydrograph at Meredosia shows a double peak. The first of 112,000 cfs occurred on
December 12, followed by another of 104,000 cfs on December 17 and 18, 1995. Had the
tributary flows occurred a few days earlier or that at Kingston Mines a few days later (Figure
46), the flood hydrograph at Meredosia would have been a single peak hydrograph with
significantly higher .
122
Figure 45. Flood hydrographs at Illinois River and tributary gaging stations, March 1985 flood
123
Figure 46. Flood hydrographs at Illinois River and tributary gaging stations, Dec. 1982 flood
124
From Marseilles to Kingston Mines, a of 87,800 cfs occurred at Marseilles on
December 4, and a of 86,700 cfs occurred at Kingston Mines on December 7. values of
22,800 and 30,000 cfs were observed in the Fox and Vermilion Rivers on December 3 and 4,
respectively. values at Marseilles and Kingston Mines are the highest in the available
record and have a lag of three days. The daily flow hydrographs at Marseilles and Kingston
Mines are single-peaked and well defined as shown in Figure 46. The moderating influence
of Peoria Lake flood surcharge storage is evident in that values for Marseilles, Dayton, and
Leonore (87,800, 22,800, and 30,000 cfs) are merged and attenuated greatly to form a single-
peaked, well-defined outflow hydrograph, which is joined by a 51,000 cfs flood peak in the
Mackinaw River near Green Valley on December 6, resulting in a of 88,800 cfs at
Kingston Mines on December 7.
4. Flood of April 1979. Relevant daily flow peak values are used for the Spoon River,
Fox River, and Illinois River at Kingston Mines and Marseilles. These flow values are
somewhat lower than corresponding values, but they provided expected lag times. A
of 109,000 cfs at Meredosia occurred on April 19, 1979. Figure 47 shows multiple-peak
hydrographs at practically all gaging stations covered. Local daily flow peaks of 101,000 cfs
occurred at Meredosia on March 24-25 and April 2-4. A relevant daily flow peak of 65,500
cfs occurred on April 15 at Kingston Mines though the of 72,200 cfs was observed on
March 24. Relevant daily flow peaks of 7460, 54,200, and 10,300 cfs occurred on April 12,
15, and 13, respectively, in the LaMoine, Sangamon, and Spoon Rivers. Flows at Kingston
Mines and in the Sangamon River are rather in phase timewise and result in a sharp
hydrograph peak at Meredosia.
The relevant daily flow peak of 62,700 cfs occurred at Marseilles on April 13 in the
Marseilles to Kingston Mines reach. Corresponding flow values of 17,000 and 16,400 cfs
occurred in the Fox and Vermilion Rivers on April 12 and 13. These flows are rather in phase
in travel along the Illinois River, resulting in a daily flow peak of 65,500 cfs on April 15 at
Kingston Mines. Peoria Lake surcharge storage delayed the peak by only a day because of the
high lake level from preceding high flows.
125
Figure 47. Flood hydrographs at Illinois River and tributary gaging station: April 1979 flood
126
5. Flood of June 1974. The of 110,000 cfs at Meredosia occurred on June 29.
The relevant daily flow peak of 69,800 cfs occurred on June 24 at Kingston Mines and values
of of 32,700 and 42,800 on June 24 and June 25 in the Spoon and Sangamon Rivers. The
relevant daily flow peak of 6,230 cfs in the La Moine River occurred on June 26 ( of 9,860
occurred on June 2). All these flow peaks appear to be in good phase timewise and resulted in
of 110,000 cfs at Meredosia (Figure 48).
From Marseilles to Kingston Mines, the relevant values and their dates of
occurrence follow: 34,600 cfs on June 23 at Marseilles, 10,400 cfs on June 23 for the Fox
River, 16,200 cfs on June 23 for the Vermilion River near Leonore, and 69,800 cfs on June 24
at Kingston Mines. Figure 48 shows multiple peak hydrographs, but the peak segments are
distinct and separate.
127
Figure 48. Flood hydrographs at Illinois River and tributary gaging stations, June 1974 flood
128
COMPARISONS AND CONCLUSIONS
The design floods and stages developed for the Dlinois River and major tributaries
below Peoria Lake are compared with similar information available from some reports and
accessory maps prepared by the Rock Island and St. Louis Districts of the U.S. Army Corps of
Engineers (USCOE). Broad conclusions and suggestions are made from the overall
information developed in this report.
T-Year Water Elevations in Alton Pool, Illinois River
There are six stage stations in the Alton Pool (river mile or RM 80.2 from LaGrange
Lock and Dam to RM 0.0 at the mouth of the Illinois River). Pertinent data for these stations
are given below.
Location RM Continuous record
niinois River at Meredosia 70.8 2/1884-12/1993
Illinois River at Valley City 61.3 1/1884-12/1993
Illinois River at Florence 56.0 1/1942-12/1993
Illinois River at Pearl 43.2 1/1942-12/1993
niinois River at Hardin 21.6 2/1932-12/1993
Mississippi River at Grafton -0.2 9/1929-12/1993
After 1940, no new dams and levees were constructed along the Illinois River.
Concurrent data at four stations (Meredosia, Valley City, Hardin, and Grafton) cover calendar
years 1941-1993. Data at other two stations (Florence and Pearl) cover 1942-1993. Values
derived for 10-, 25-, 50-, 100-, and 500-year water elevations at the six stations, using the
general frequency program, are given in Table 27. Values of two maximum observed
elevations at the gaging stations and their fitted recurrence interval (using mixed distribution)
in years are also included. Values of 10- to 500-year water elevations were read from Plate A-
38, Water Surface Profiles, Mouth to Mile 80.0, report from the USCOE, St. Louis District,
1981. Due to the absence of water elevation data for the exceptionally high floods and stages
in July or August 1993 in the Mississippi River in the 1981 study, the USCOE water
129
Table 27. Comparison of T-year Water Elevations in this Study and from USCOE Report for Alton Pool
Obsd two River Record Design WE, feet-msl, for T equal to max WE Fitted T,
Illinois River at mile Source used 10 25 50 100 500 (feet-msl) Year years
Meredosia 70.8 Table 13 1941-1993 442.92 445.38 447.08 448.59 451.89 446.69 1943 43 USCOE 442.50 444.95 446.00 446.80 448.25 445.60 1985 27
A 0.42 0.43 1.08 1.79 3.64
Valley City 61.3 Table 13 1941-1993 441.22 443.84 445.61 447.29 450.88 444.91 1943 38 USCOE 440.65 443.10 444.50 445.50 447.00 444.10 1993 28
A 0.57 0.74 1.11 1.79 3.88
Florence 56.0 Table 13 1942-1993 440.19 442.85 444.66 446.35 449.97 443.60 1993 33 USCOE 439.50 441.95 443.35 444.50 446.20 442.80 1943 25
A 0.69 0.90 1.31 1.85 3.77
Pearl 43.2 Table 13 1942-1993 437.44 440.15 442.07 443.91 448.01 442.72 1993 64 USCOE 437.10 439.60 441.25 442.60 444.50 440.47 1973 28
A 0.34 0.55 0.82 1.31 3.51
Hardin 21.6 Table 13 1941-1993 434.55 437.49 439.53 441.47 445.71 442.30 1993 136 USCOE 434.30 436.50 438.15 440.20 443.90 438.20 1973 32
A 0.25 0.99 1.38 1.27 1.81
Mississippi at Grafton -0.2 Table 13 1941-1993 432.52 435.81 438.09 440.24 444.86 441.80 1993 193 USCOE 432.40 435.30 437.50 439.85 443.70 436.89 1973 31
A 0.12 0.51 0.59 0.39 1.16
Notes: Table 13 is in this report. USCOE refers to Rock Island District, Plate A-38, Water Surface Profiles, May 1981, fitted T-year values to two maximum observed stages are with the mixed distribution, and A in feet denotes WE from Table 13 minus WE from USCOE.
elevations are lower than those derived in this study. The differences, A values (water
elevation from Table 13 minus water elevation from USCOE) are included in Table 27.
These A values become significantly and progressively higher with increase in T for stage
gaging stations at Meredosia, Valley City, Florence, and Pearl.
T-Year Flood Peaks in Major Tributaries
There are five major tributaries to the Illinois River downstream of Peoria Lock and
Dam.
RM at Gaging station Continuous record Tributary mouth u/s of mouth (mi) (water years)
Mackinaw River near Green Valley 147.7 13.7 1922-1993
Spoon River at Seville 120.4 38.7 1919-1993
Sangamon River near Oakford 88.9 25.7 1931-1993
La Moine River at Ripley 83.5 12.3 1921-1993
Macoupin Greek near Kane 23.2 16.1 1941-1993
The 10-, 25-, 50-, 100-, and 500-year design floods developed in this study for these
tributaries, using the general frequency method and consideration of any increasing trend in
flood peaks, are given in Table 28. Corresponding flood peaks (USCOE, 1992), available for
four stations excluding Macoupin Creek near Kane, are also given in Table 28. The major
differences in T-year flood peaks are mostly in Mackinaw River near Green Valley for T = 10
to 100-years.
T-Year Flood Peaks at Illinois River Gaging Stations
There are three gaging stations on the Illinois River used for flood frequency analyses
in this report.
The 10-, 25-, 50-, 100-, and 500-year design floods developed in this study at these
gaging stations, using the general frequency method and consideration of any increasing trend
in flood peaks, are given in Table 29. Corresponding flood peaks (USCOE, 1992) are also
given in the table. The T-year flood peaks at Kingston Mines and Meredosia are not much
131
Table 28. Comparison of T-year Floods in Major Tributaries to the Illinois River
T-year flood peaks River and Station Source 10 25 50 100 500
Mackinaw River Table 16 27,600 39,600 50,000 62,000 97,800 USCOE* 21,100 31,300 40,800 52,200 88,200 USCOE† 21,500 32,500 43,100 56,300 101,000
Spoon River Table 16 26,100 32,300 37,800 41,000 51,800 USCOE* 23,600 29,800 34,500 39,400 51,100 USCOE† 23,900 30,300 35,400 40,700 54,100
Sangamon River Table 16 46,300 59,000 72,000 87,500 138,000 USCOE* 49,700 64,500 75,500 86,500 112,000 USCOEf 50,300 65,800 77,700 89,700 118,000
LaMoine River Table 16 23,000 26,800 31,000 35,000 45,000 USCOE* 17,500 22,400 26,200 30,300 40,500 USCOEf 17,700 22,900 27,100 31,600 43,400
Macoupin Creek Table 16 22,800 29,300 34,600 39,600 59,400
Notes: Table 16 is in this report. USCOE refers to Rock Island District, Illinois River Water Surface Profiles, 1992 report. * denotes computed flow and t denotes modified flow used in modeling.
132
different. However, the values at Marseilles as derived in this study are significantly higher
than those of USCOE, mostly because the latter used flood data before 1941 with greater
proportion of low flood peaks.
River Drainage area Continuous record Illinois River/station mile (mi) (sq mi) (water years)
Illinois River at Marseilles 246.5 8,259 1920-1993
Illinois River at Kingston Mines 144.4 15,818 1941-1993
niinois River at Meredosia 7.1.3 26,028 1939-1993*
Note: * denotes flow data at Valley City gaged adjusted to that at Meredosia for years 1990-1993
T-Year Water Elevations at Marseilles, Kingston Mines, and Meredosia
According to the U.S. Army Corps of Engineers (USCOE), Rock Island District, no
stage or water elevation frequency analyses were conducted with the observed maximum
annual stages or water elevations at these three stations. The unsteady flow model profiles
yield the approximate relevant T-year water elevations and the same are given in Table 30.
The 10-, 25-, 50-, 100-, and 500-year design water elevations, developed in this study
using the general frequency method and consideration of any increasing trend in water
elevations, are taken from in Table 23 of this report and given in Table 30. These values at
Meredosia are somewhat higher than the corresponding values in Table 13 and Table 27
because the latter were developed from 1941-1993 data, without consideration of any
increasing trend.
Difference in water elevations (Table 23 values minus those from USCOE water
surface profiles), or A feet, are also given in Table 30. At Marseilles, the USCOE values are
2.67 (for T = 10) to 1.35 (T = 500) feet lower than corresponding values developed in this
study. USCOE values are significantly lower, probably due to using data including years
1920-1940 which had a higher proportion of lower water elevations. At Kingston Mines,
values of A are rather low, starting with 0.10 feet (T = 10) to 0.70 feet (T = 500). At
Meredosia, the corresponding range is -0.18 to 1.16.
133
Table 29. Comparison of T-year Floods at Marseilles, Kingston Mines, and Meredosia
T-year flood peaks Illinois River station Source 10 25 50 100 500
at Marseilles Table 20 84,200 96,000 105,000 113,000 131,000 USCOE 73,600 87,400 97,300 107,000 128,000
at Kingston Mines Table 20 76,500 85,000 91,800 98200 113,000 USCOE 74,000 85,600 93,600 101000 118,000
at Meredosia Table 20 105,000 120,600 130,000 139,500 161,000 USCOE 100,000 117,000 129,000 140,000 164,000
Notes: Table 20 is in this report. USCOE refers to Rock Island District, Illinois River Water
Surface Profiles, 1992 report.
Table 30. Comparison of T-year Water Elevations at Marseilles, Kingston Mines, and Meredosia
T-year water elevations, feet-msl and ∆ feet Illinois River station Source 10 25 50 100 500
at Marseilles Table 23 477.87 479.31 480.43 481.16 483.05 USCOE 475.2 476.7 478.5 479.5 481.7 ∆, feet 2.67 2.61 1.93 1.66 1.35
at Kingston Mines Table 23 451.40 453.50 454.6 455.8 459.10 USCOE 451.3 453.2 454.6 455.8 458.4 ∆, feet 0.10 0.30 0.30 0.25 0.70
at Meredosia Table 23 443.82 446.38 448.08 449.55 452.76 USCOE 444.0 446.5 447.9 449.1 451.6 ∆, feet -0.18 -0.12 0.29 0.45 1.16
Notes: Table 23 is in this report. USCOE refers to Rock Island District, Illinois River Water Surface Profiles, 1992 report. A in feet = difference between Table 23 and USCOE values.
134
Conclusions
The general frequency program is very versatile for fitting probability distributions to
observed maximum floods and stages, which indicate fitting curves to be straight lines or with
curvatures of various shapes when plotted on lognormal probability paper. It also provides an
objective way of detecting outliers/inliers at both the low and high end of the flood spectrum
and their modification at various significance levels as desired.
The annual maximum water elevations at six stations in the Alton Pool of the Dlinois
River are largely governed by the relative severity of the Mississippi backwaters and the
magnitude of flood peak and volume at Meredosia. During the 1941-1993 period, the highest
water elevations were recorded in 1993 at Grafton, Hardin, Pearl, and Florence, but at
Florence the rank was 2 and at Meredosia, the rank was 4 (the flood at Meredosia was only a
4- or 5-year flood). However, the highest water elevations were recorded at Valley City and
Meredosia in 1943 when the flood peak at Meredosia was the highest observed. Flood water
surface profiles in the Alton Pool depend on the joint probability of high Mississippi
backwaters and high Meredosia flood peaks.
Flood frequency analyses of the annual floods observed in five major tributaries
(Mackinaw River, Spoon River, Sangamon River, La Moine River, and Macoupin Creek)
show that there is 1) a significant trend of increase in values of both high and low flood over
time in Mackinaw River and La Moine River, 2) no significant trend for the Spoon River and
Sangamon River, and 3) a slight decreasing trend in value of high floods and a significant
increasing trend in value of low floods in the Macoupin Creek.
For the Illinois River at Marseilles, Kingston Mines, and Marseilles, the trend of
increase in high floods is significant at all three stations. A similar trend is exhibited by the
observed high stages. Consideration of these trends has been incorporated in developing
design T-year floods, stages, and water elevations.
Ten highest floods and their daily flow hydrographs observed at Meredosia (1941-
1993 period) were analyzed in terms of floods and hydrographs associated with that observed
at Meredosia for Illinois River at Marseilles and Kingston Mines and major tributaries: Fox
River, Vermilion River, Mackinaw River, Spoon River, Sangamon River, La Moine River,
135
and Macoupin Creek. For the Illinois River from Kingston Mines to Meredosia and the major
tributaries in this reach and Mackinaw River, the relevant flood ranks (1 is the highest) are
given below.
Kingston Month Year Meredosia La Moine Sangamon Spoon Mines Mackinaw
11
10
11
6
30
8
2
3
1
6
7
4
28
3
9
30
1
17
1
21
2
3
6
29
18
1
6
34
29
28
1
2
3
4
5
6
May 1943
Feb-Mar 1985
Dec 1982
April 1979
June 1974
March 1982
There seems to be better correlation between flood ranks at Meredosia and Kingston
Mines. The relative efficiency drops to 2/3 for Sangamon, to 1/3 for La Moine and Spoon,
and to 1/6 for Mackinaw River. While working for the U.S. Army Corps of Engineers, Robert
Barkau used the December 1982 flood to verify the UNET model results and to develop T-
year flood profiles by using flow adjustment factors (considering the 1982 flood as a 25-year
flood) for river reaches from Lockport to Marseilles, Marseilles to Kingston Mines, and
Kingston Mines to Meredosia. The use of adjustment factors implies relative tributary flood
contributions for various recurrence intervals to be proportionate to those for the December
1982 flood. This assumption neglects higher flood stages occurring in portions of reaches
depending on variability in the tributary flood. A new scheme needs to be developed for
better simulation of flood profiles under varying flood contributions from the tributaries.
136
REFERENCES Barkau, R.L. 1993. UNET -- One-Dimensional Unsteady Flow Model Through a Full
Network of Open Channels: User's Manual. U.S. Army Corps of Engineers, Hydrologic Engineering Center, Davis, CA.
Barnett, V., and T. Lewis. 1978. Outliers in Statistical Data. John Wiley and Sons, New York, NY, 356 p.
Blom, G. 1958. Statistical Estimates and Transformed Beta Variables. John Wiley and Sons, New York, NY, 176 p.
Box, G.E.P., and D.R. Cox. 1964. An Analysis of Transformation. Journal of Royal Statistical Society, Series B, Vol. 26, pp. 211-252.
Box, G.E.P., and C.G. Tiao. 1973. Bayesian Inference in Statistical Analysis. Addison-Wesley, Reading, MA, 588 p.
Canterford, R.P., and C.L. Pierrehumbert. 1977. Frequency Distributions for Heavy Rainfalls in Tropical Australia. In Hydrology Symposium, Institute of Civil Engineers, Brisbane, Australia.
Cohen, A.C. 1967. Estimation in Mixtures of Normal Distributions. Technometrics, Vol. 9, No. 1, pp. 15-28.
Hald, A. 1952. A Statistical Theory with Engineering Applications. John Wiley and Sons, New York, NY, pp. 152-154.
Kite, G.W. 1977. Frequency and Risk Analyses in Hydrology. Water Resources Publications, Fort Collins, CO, p. 46.
Moran, P.A.P. 1959. The Theory of Storage. John Wiley and Sons, New York, NY, p. 14.
Singh, K.P. 1968. Hydrologic Distributions Resulting from Mixed Populations and Their Computer Simulation. International Association of Scientific Hydrology, Tucson Symposium, pp. 671-681.
Singh, K.P. 1974. Two-Distribution Method for Fitting Mixed Distributions in Hydrology. In Statistical Hydrology Symposium, Miscellaneous Publication 1275, U.S. Department of Agriculture, pp. 371-382.
Singh, K.P. 1980a. Discussion on — Flood Frequency Analysis by Power Transformation. ASCE, Journal of Hydraulics Division, Vol. 106, No. HY3, pp. 462-465.
Singh, K.P. 1980b. Flood Frequency Modeling and Outliers. Proceedings, ASCE Computer and Physical Modeling in Hydraulic Engineering Conference, pp. 74-85.
137
Singh, K.P. 1983. Software Design for a Versatile Flood Frequency Analysis. Advances in Engineering Software, Vol. 5, No. 2, pp. 107-112.
Singh, K.P. 1987. Development of a Versatile Flood Frequency Methodology and Its Application to Flood Series from Different Countries. In Hydrologic Frequency Modeling, D. Reidel Publishing Company, Boston, MA, pp. 183-197.
Singh, K.P. 1991. Illinois River: Old Issues, Emerging Issues, and Management. Paper presented at the 1991 Conference of the American Water Resources Association, Illinois Section.
Singh, K.P., and M. Nakashima. 1981. A New Methodology for Flood Frequency Analysis with Objective Detection of Outliers/Inliers. Illinois State Water Survey Contract Report 272, 145 pages.
Singh, K.P., and G.S. Ramamurthy. 1990. Changes in Climate and Hydrology of the Illinois River Basin and What They Portend for Levee Districts and Peoria Lake. Proceedings, American Water Resources Association Conference held in Peoria, Illinois, pp. 29-38.
Singh, K.P., and R.A. Sinclair. 1972. Two-Distribution Method for Flood Frequency Analysis. ASCE, Journal of Hydraulics Division, Vol. 98, No. HY1, pp. 29-44.
United States Army Corps of Engineers (USCOE). 1982. Illinois River Water Surface Profiles River Mile 80 to 290 Unsteady Flow Model. Rock Island District.
U.S. Water Resources Council. 1973. Guidelines for Determining Flood Flow Frequency. Bulletin
17A. Washington, DC. Waylen, P., and M.K. Woo. 1982. Prediction of Annual Floods Generated by Mixed
Processes. Water Resources Research, Vol. 18, No. 4, pp. 1283-1286.
138